Copyright TechTarget - All rights reserved https://cyber.law.harvard.edu/rss/rss.html Techtarget Feed Generator en Mon, 07 Sep 2026 18:20:31 GMT https://www.techtarget.com/searchenterpriseai editor@techtarget.com <p>Contact centers have been an effective way to take advantage of AI advancements. These technologies deliver businesses rapid ROI and actionable insights that can streamline processes and improve operational efficiency.</p> <p>AI now accurately and conveniently resolves customer issues across several communication channels using voice, text, messaging and other emerging channels. Additionally, businesses can take advantage of improved contact center visibility and predictive insight through&nbsp;AI-derived analytics, metrics and KPIs.</p> <section class="section main-article-chapter" data-menu-title="3 main types of contact center AI"> <h2 class="section-title"><i class="icon" data-icon="1"></i>3 main types of contact center AI</h2> <p>The first category of AI is <a href="https://www.techtarget.com/searchenterpriseai/definition/conversational-AI">conversational</a> and <a href="https://www.techtarget.com/ai/feature/How-generative-AI-is-changing-creative-work">generative AI</a>, which uses large language models (<a href="https://www.techtarget.com/whatis/definition/large-language-model-LLM">LLM</a>s) combined with Retrieval-Augmented Generation (RAG) and natural language understanding. These technologies enable natural conversations across voice, text, messaging and other channels through modern interactive voice response (<a href="https://www.techtarget.com/searchcustomerexperience/definition/Interactive-Voice-Response-IVR">IVR</a>) systems, chatbots and virtual assistants.</p> <p>A closely related and fast-growing category in the contact center space is agentic AI. These systems go beyond basic conversations and push to execute multi-step actions. This allows contact center systems to work autonomously, only escalating to human agents when absolutely needed.</p> <p>The third type of contact center AI&nbsp;uses <a href="https://www.techtarget.com/data-technologies/tip/Generative-AI-can-improve-not-replace-predictive-analytics">predictive generative analysis</a> to process interaction data, statistics and KPIs with the goal of making recommendations&nbsp;on how to improve operational performance or increase customer satisfaction. This type of AI helps contact center operators meet their performance goals without having to manually sift through and analyze data using manual or semi-automated processes.</p> </section> <section class="section main-article-chapter" data-menu-title="5 popular contact center AI features"> <h2 class="section-title"><i class="icon" data-icon="1"></i>5 popular contact center AI features</h2> <p>IVR systems, voice AI, chatbots, agentic virtual agents,&nbsp;agent coaching and monitoring, predictive analytics and generative AI capabilities are some of the key features in a modern contact center. AI-generated channel summaries are among the more popular, most widely adopted and high-impact capabilities in modern contact center platforms.</p> <h3>1. Advanced conversational AI</h3> <p>Traditional, menu-based IVR has rapidly evolved due to AI. Modern conversational AI now powers natural voice interactions with support for multiple languages and use advanced biometrics capabilities for contact authentication. These conversational AI systems can complete routine tasks, retrieve information from an LLM in real time and transfer callers to the appropriate human agent when required. When properly applied, organizations can expect shorter wait times and a smoother customer experience.</p> <h3>2. Agentic self-service chatbots and virtual agents</h3> <p>Modern virtual agents now go beyond scripted chatbots. Powered by custom LLMs, RAG and integrated tools, self-service chatbots and virtual agents now understand and can assist with the following workflows:</p> <ul class="default-list"> <li>Handle complex customer requests with accurate, context-aware responses.</li> <li>Retrieve and act on live data from CRM and other enterprise systems.</li> <li>Execute multi-step transactions across connected tools.</li> <li>Resolve issues end to end without human intervention when appropriate.</li> </ul> <p>When escalation to a human agent is needed, AI chatbots and agents deliver a full conversation and task context history. This helps drive faster resolution rates, delivers a consistent experience and can operate 24/7 with minimal staffing.</p> <h3>3. Real-time agent coaching and performance monitoring</h3> <p>At this stage, most contact centers still use a combination of AI-powered IVR, chatbots, virtual assistants and human agents. For the human side of the operation, AI continues to improve the customer service experience. Nearly every aspect of a human agent's contact with customers can be analyzed in real-time using AI.</p> <p><a href="https://www.techtarget.com/searchcustomerexperience/tip/Top-7-call-center-agent-performance-metrics-to-track">Examples of collected metrics</a>&nbsp;include call and chat logs, handle times, time-to-service resolution, queue times, hold times and customer survey results. All this information is collected and analyzed to determine how customer satisfaction can increase, while simultaneously decreasing time-to-service resolution. AI is used to track these statistics, formulate performance profiles and&nbsp;<a href="https://www.techtarget.com/searchcustomerexperience/answer/Nine-skills-every-call-center-agent-job-requires">make automated coaching suggestions to agents</a>.</p> <h3>4. Automatic call insights with predictive analytics</h3> <p>CRM&nbsp;software is commonly used with contact center platforms and stores a wealth of customer data, including contact info, purchase preferences and history and previous interaction touchpoints. AI now commonly combines CRM data with real-time interaction signals to intelligently deliver relevant context and predictive recommendations to both human and virtual agents. These capabilities also support proactive engagement, helping contact center managers anticipate needs, improve personalization and expand revenue upsell opportunities.</p> <h3>5. AI-generated transcription, call and chat summaries</h3> <p>Generative AI and advanced language models are now used to transcribe, organize and summarize post-call and post-chat summaries. These rich summaries can then be put into a CRM system and further analyzed to determine various aspects of a customer's interaction with the contact center, including their overall satisfaction, likelihood of purchasing products and services in the future, brand&nbsp;<a target="_blank" href="https://www.shopify.com/blog/loyalty-program" rel="noopener">loyalty</a> and which targeted marketing and sales methods are most likely to translate into future sales.</p> </section> <section class="section main-article-chapter" data-menu-title="Choosing the right contact center AI platform"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Choosing the right contact center AI platform</h2> <p>If you're evaluating contact center platforms and are interested in understanding what to look for when it comes to AI, focus on the following three criteria:</p> <h3>1. Strong integration with existing apps and tools</h3> <p>Contact center operations include a number of tools where AI platforms must integrate cleanly. This includes CRM platforms, knowledge bases, workforce management, business intelligence (BI) ticketing and underlying telephony infrastructure. Look for a contact center platform with API and <a href="https://www.techtarget.com/ai/tip/How-the-Model-Context-Protocol-simplifies-AI-development">Model Context Protocol (MCP) support</a>.</p> <h3>2. Mature generative and agentic AI features</h3> <p>Identify tools with proven support for LLM, RAG and the ability for AI to complete multi-step actions fully autonomously. Also be sure that platforms have clear guardrails and reliable escalation paths to human agents so AI can operate safely and reliably inside a contact center environment.</p> <h3>3. Clear and measurable outcomes</h3> <p>When evaluating contact center platforms, identify those that deliver clear, actionable metrics that managers and human agents can use to track and improve over time. The strongest platforms for your environment will highlight performance data on resolution rates, handle times, customer satisfaction and others. AI should be able to turn these insights into real-time coaching, automated quality feedback checks and other improvement opportunities for all contact center agents and managers.</p> <p><em>Andrew Froehlich is founder of InfraMomentum, an enterprise IT research and analyst firm, and president of West Gate Networks, an IT consulting company. He has been involved in enterprise IT for more than 20 years.</em></p> </section> As multifunctional contact centers grow more complex and vital as revenue and relationship centers, generative and agentic AI have become a foundational component. https://cdn.ttgtmedia.com/rms/onlineimages/chatbot_g1250576673.jpg https://www.techtarget.com/enterprise-software/feature/5-key-contact-center-AI-features-and-their-benefits Thu, 03 Sep 2026 09:00:00 GMT <p>Boomi is betting big on a control layer for AI.</p> <p>The vendor, which began as <a href="https://www.techtarget.com/data-technologies/news/366554135/New-Boomi-AI-tool-enables-natural-language-data-integration">a data integration specialist</a> but now focuses on enabling customers to activate their data for AI, on Tuesday launched the Agent Control Plane. The new layer brings previously disparate Boomi capabilities together in a single environment and is designed to provide customers with capabilities that enable them to connect AI agents with business systems, govern agents in production and <a href="https://www.techtarget.com/enterprise-software/feature/AI-feature-spend-is-the-new-software-cost-control-problem">rein in spending on AI</a>.</p> <p>Keeping control over agents and the cost of doing so are rising concerns for many enterprises as they try to <a href="https://www.deloitte.com/content/dam/assets-zone3/us/en/docs/services/consulting/2026/state-of-ai-2026.pdf">move past experiments with AI</a> and put agents into production. Boomi's Agent Control Plane is therefore a valuable addition for the vendor's users, according to Michael Ni, an analyst at Constellation Research.</p> <p>"Boomi is making a bid for one of the more valuable control points in enterprise AI, the layer between agents and the systems where business actually gets done," he told TechTarget. "As companies add more models and agents, they need a way to control what those agents can access, what they can do and what they cost."</p> <p>Boomi is not the only vendor adding tools aimed at giving customers better control over their agents, Ni continued, noting that AWS, Salesforce and ServiceNow are among the others that also provide features for orchestrating agents. However, Boomi's integration capabilities give it a foundation from which to build <a href="https://www.techtarget.com/it-strategy/feature/The-AI-agent-governance-gap-How-CIOs-can-gain-control">a control plane for agents</a> that could help it compete for market share.</p> <p>"The idea of an Agent Control Plane is quickly becoming a crowded but strategic land grab," Ni said. "Boomi's advantage is … to turn its footprint into a neutral data activation layer where enterprise data can be contextualized, governed and acted upon across models and platforms."</p> <section class="section main-article-chapter" data-menu-title="Control vs. chaos"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Control vs. chaos</h2> <p>Agents that are improperly overseen can cause serious problems.</p> <p>For example, an agent powered by Anthropic's Claude large language model <a target="_blank" href="https://www.theguardian.com/technology/2026/apr/29/claude-ai-deletes-firm-database" rel="noopener">deleted a company's entire production database</a> last April. Two months earlier, another agent allegedly <a target="_blank" href="https://sumsub.com/media/news/developer-warns-ai-agent-defamation-post-shows-risks-of-autonomous-ai/" rel="noopener">wrote a blog post</a> defaming a developer after the developer rejected the agent's code contribution. Beyond headline-grabbing mishaps, agents that aren't closely controlled can drive up an organization's spending by getting caught in recursive loops and consuming unlimited amounts of tokens.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Boomi is making a bid for one of the more valuable control points in enterprise AI, the layer between agents and the systems where business actually gets done. As companies add more models and agents, they need a way to control what those agents can access, what they can do and what they cost. </figure> <figcaption> <strong>Michael Ni</strong>Analyst, Constellation Research </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>Gartner <a target="_blank" href="https://www.gartner.com/en/newsroom/press-releases/2026-05-26-gartner-says-applying-uniform-governance-across-ai-agents-will-lead-to-enterprise-ai-agent-failure" rel="noopener">predicts</a> that by the start of 2027, 40% of organizations will demote or decommission agents due to a lack of proper governance.</p> <p>Boomi's Agent Control Plane is designed to provide the governance enterprises need to trust that their AI tools appropriately connect to enterprise systems such as Salesforce and Workday, act only as intended in production, and do so within an expected spending range.</p> <p>As a result, though other vendors are similarly addressing AI governance, Boomi's control layer for agents is significant for its customers, according to Stephen Catanzano, an analyst at Omdia, a division of TechTarget.</p> <p>"The Agent Control Plane represents a big addition because it addresses what Boomi identifies as the largest unmanaged risk in enterprise technology today, [which is] giving AI agents the deep access they need to transactional systems while maintaining governance, cost control, and security that most organizations currently lack," he told TechTarget.</p> <p>Meanwhile, there is some potential differentiation between Boomi's approach to governing AI and competing capabilities from Salesforce through <a href="https://www.computerweekly.com/blog/CW-Developer-Network/MuleSoft-Omni-Gateway-offers-route-to-agentic-visibility-consistent-governance">MuleSoft</a>, AWS and specialists such as Apigee, Catanzano continued.</p> <p>"Boomi seems to be betting that their unified approach sitting between any agent and any system, rather than being tied to a single cloud ecosystem, provides the differentiation," he said.</p> <p>Boomi's Agent Control Plane is designed to be flexible, running across public clouds, customers' virtual private clouds and on-premises environments. In addition, it supports data and <a href="https://www.techtarget.com/cybersecurity/feature/The-push-for-digital-sovereignty-What-CISOs-need-to-know">digital sovereignty</a> to ensure that customers remain compliant with the rules and regulations where their information is created.</p> <p>The Agent Control Plane combines pre-existing Boomi capabilities to provide the following:</p> <ul class="default-list"> <li>A single, centralized location for overseeing agents and tools across platforms, including putting limits on <a href="https://www.techtarget.com/whatis/definition/token">token</a> consumption and stopping high-risk actions until they have human approval.</li> <li>Enforcement through the Boomi AI Gateway, which combines MCP Gateway and LLM Gateway in one control panel and is built on the technology the vendor acquired when <a href="https://www.constellationr.com/insights/news/boomi-buy-lunardev-eyes-intelligent-prompt-and-model-routing">it bought Lunar.dev</a>.</li> <li>Governed connectivity through Boomi Connect MCP servers so that agents can safely call on enterprise systems to inform their actions.</li> <li>Deployment flexibility via a hybrid runtime architecture to adhere to data sovereignty rules, secure sensitive information and control spending.</li> <li>Data semantics and lineage to ground agent reasoning with approved data.</li> <li>Natural language processing capabilities in Boomi Orchestrate so that non-technical teams can build agentic AI workflows.</li> </ul> <p>"This came directly from what we were hearing in a critical mass of customer conversations, not from a single feature request," Ed Macosky, Boomi's chief product and technology officer, told TechTarget.</p> <p>User feedback consistently included concerns over <a href="https://www.computerweekly.com/news/366645085/Boomi-CEO-shares-vision-of-AI-cost-management">increasing spending on AI</a>, difficulty connecting agents with disparate transactional systems and fear of losing control over intellectual property once it's accessed by agents, he continued.</p> <p>"The Agent Control Plane exists because enterprises shouldn't have to choose between moving fast on AI and keeping control of their data and infrastructure," Macosky said.&nbsp;</p> <p>Meanwhile, Boomi's control layer for AI appears logically constructed by combining centralized visibility with enforcement capabilities, deployment flexibility and <a href="https://www.techtarget.com/data-technologies/opinion/Tracing-data-lineage-in-AI-systems">data lineage</a>, according to Catanzano.</p> <p>"The architecture directly maps to the three problems Boomi highlights -- governance gaps that cause production incidents, runaway token costs that have made AI spend management the most in-demand FinOps skill, and trust deficits," he said.</p> <p>Ni likewise noted that the Agent Control Plane features the capabilities needed to provide Boomi's customers with a single environment for <a href="https://www.techtarget.com/data-technologies/feature/Data-and-AI-governance-must-team-up-for-AI-to-succeed">governing AI</a>. In addition, it does so by combining tools that have already been proven to work.</p> <p>"Boomi has put the major pieces together and built on areas where it already has known strengths," Ni said. "Boomi's next step will be to go deeper into runtime context and managing agents across their full lifecycle. ... That becomes increasingly important as enterprises move from experimenting with agents to operating them at scale."</p> </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Looking ahead</h2> <p>After launching the Agent Control Plane , Macosky said that Boomi's product development plans for the rest of 2026 focus on improving AI governance across providers and models, broadening its partnership network to add openness, and enhancing basic platform features such as <a href="https://www.techtarget.com/data-technologies/feature/Why-AI-forces-security-first-governance">security</a> and reliability.</p> <p>One specific way the vendor could better serve its customers would be to add tools that show whether agents in action delivered their desired outcome, according to Ni.</p> <p>"Boomi already connects the data, exposes the tools, governs the agent and orchestrates the action. Now it needs to show whether that action delivered the desired business outcome,&nbsp;then&nbsp;use that feedback to improve the next decision," he said. "That would&nbsp;make Boomi’s positioning around data activation more tangible to CIOs and business leaders."</p> <p>Catanzano similarly suggested that Boomi expand beyond governance and connectivity to measure and improve the <a href="https://www.techtarget.com/enterprise-software/feature/Software-value-needs-outcome-measures">business outcomes</a> generated by agents.</p> <p>"Boomi could continue serving users and attract new ones by … helping organizations not just control what their agents do, but systematically improve how well they do it through analytics that connect agent actions to actual business metrics," he said.</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> The Agent Control Plane combines previously disparate capabilities in one environment to help users better observe and command their fleets of agents. https://cdn.ttgtmedia.com/rms/onlineimages/machine_learning_g1303163039.jpg https://www.techtarget.com/data-technologies/news/366649690/Boomi-intros-layer-for-controlling-AI-costs-and-connections Wed, 02 Sep 2026 09:49:00 GMT <p>IT automation uses instructions to create clear, consistent and repeatable processes that replace an IT professional's manual work in data centers and cloud deployments. The scope of IT automation ranges from single actions to discrete sequences and autonomous IT deployments that take actions based on administrative requirements, user activities and other event triggers.</p> <p>At the technical level, IT automation enables administrators to complete time-consuming, resource-intensive and error-prone tasks faster and more consistently with minimal human intervention. At the operational level, it helps organizations standardize processes, improve documentation and meet security, regulatory and <a href="https://www.techtarget.com/searchitoperations/definition/compliance-automation">compliance</a> requirements.</p> <p>This guide examines many aspects of IT automation, including benefits, challenges, technologies and trends. Readers will learn what IT teams must do to build a business case for automation, integrate AI-infused tools into their processes and achieve <a href="https://www.techtarget.com/searchitoperations/definition/What-is-workflow-orchestration">workflow orchestration</a> on a grand scale.</p> <section class="section main-article-chapter" data-menu-title="How IT automation works"> <h2 class="section-title"><i class="icon" data-icon="1"></i>How IT automation works</h2> <p>IT automation relies on properly installed and configured software tools to define and execute a prescribed series of detailed actions that are invoked manually or by an external trigger, such as a change in IT capacity demand.</p> <p>IT automation replaces a series of actions and responses between an administrator and the IT environment. For example, an IT automation platform such as PowerShell combines cmdlets, variables and other components into a script. This script mimics the series of commands and steps that an administrator would implement through the CLI to provision a VM or to create a backup process. An administrator can achieve a more complex IT automation outcome by combining multiple scripts into a series. These limited-scope automation processes are most beneficial when they replace a task that an administrator must perform frequently, thereby eliminating many common errors.</p> <p>IT automation is sometimes used interchangeably with the term <a href="https://www.techtarget.com/searchitoperations/definition/orchestration"><i>orchestration</i></a>, but each refers to different functions. Automation accomplishes a task repeatedly without human intervention. <a href="https://www.techtarget.com/searchitoperations/tip/IT-automation-vs-orchestration-Key-differences">Orchestration is a broader concept</a> in which the user coordinates automated tasks into a cohesive IT and business process or workflow.</p> <figure class="main-article-image full-col" data-img-fullsize="https://www.techtarget.com/rms/onlineimages/comparing_automation_and_orchestration_how_they_work-f.png "> <img data-src="https://www.techtarget.com/rms/onlineimages/comparing_automation_and_orchestration_how_they_work-f_mobile.png " class="lazy" data-srcset="https://www.techtarget.com/rms/onlineimages/comparing_automation_and_orchestration_how_they_work-f_mobile.png 960w,https://www.techtarget.com/rms/onlineimages/comparing_automation_and_orchestration_how_they_work-f.png 1280w" alt="Graphic listing the differences between automation and orchestration." height="380" width="560"> <figcaption> <i class="icon pictures" data-icon="z"></i>Automation and orchestration fulfill very different functions in business operations. </figcaption> <div class="main-article-image-enlarge"> <i class="icon" data-icon="w"></i> </div> </figure> <p>Enterprise-class IT <a href="https://www.techtarget.com/searchitoperations/definition/What-is-infrastructure-automation">infrastructure automation</a> tools trigger actions based on thresholds and other situational conditions within the IT environment. Advanced IT automation tools oversee the configuration of systems, software and other infrastructure components; recognize unauthorized or unexpected changes; and automatically take corrective actions. If a workload stops responding, automated steps kick in to restart it on a different server with available capacity. When IT automation is set to enforce a desired state of configurations, the tool detects changes in a server's configuration that are out of spec and restores it to the correct settings.</p> <p>From a practical perspective, IT automation involves four broad phases:</p> <ol type="1" start="1" class="default-list"> <li><b>Analysis.</b> IT administrators and other stakeholders assess the manual task or process to be automated.</li> <li><b>Implementation.</b> The task is translated into a series of instructions, such as scripts, workflows or other automation elements. A task, for instance, might be translated into a series of cmdlets for use in PowerShell.</li> <li><b>Integration.</b> Once tested and validated, the automation element can be integrated into the automation platform for operational use.</li> <li><b>Maintenance.</b> Automation elements must be reviewed and updated regularly, often through version control practices similar to those used in software development.</li> </ol> </section> <section class="section main-article-chapter" data-menu-title="What IT automation is used for"> <h2 class="section-title"><i class="icon" data-icon="1"></i>What IT automation is used for</h2> <p>IT operations managers and teams can <a href="https://www.techtarget.com/searchitoperations/tip/Tasks-to-automate-today-to-streamline-IT-operations">use IT automation for several tasks</a>, including the following:</p> <ul type="disc" class="default-list"> <li><b>Resource provisioning.</b> Some of the routine provisioning exercises that IT administrators must accomplish include deploying new virtualized servers, creating storage volumes and connecting networks. IT automation can significantly accelerate most of these provisioning tasks, enabling new IT environments to be implemented quickly and repeatedly.</li> <li><b>Cloud provisioning.</b> Like resource provisioning, IT automation can be used to provision CloudOps resources for workload deployments and other outcomes, such as cloud billing or report generation. As with traditional resource provisioning, automation ensures repeatable outcomes every time with faster results and fewer human errors.</li> <li><b>System configuration.</b> Often considered a provisioning task, configuration can ensure a provisioned resource is prepared and set up to accommodate application, business, infrastructure and security needs. Configuration automation is also a common aspect of IT change management strategies.</li> <li><b>Infrastructure management.</b> Countless tasks are used to manage an IT infrastructure -- from regular data backups and data retention to system and application reporting -- to ensure important management tasks are executed with the same consistency and predictability as other IT tasks.</li> <li><b>Incident management.</b> Using automation to respond to major incidents helps enterprises restore service faster and with fewer errors. IT automation helps businesses reduce the duration and cost of such incidents for themselves and their customers. An incident management ticket in response to an outage, for example, can be quickly created and assigned to the appropriate person or queue through automation.</li> <li><b>Application deployment.</b> Whether businesses use traditional or continuous integration and continuous application deployment approaches, automating essential tasks and capabilities -- particularly during testing -- can help them successfully <a href="https://www.techtarget.com/searchitoperations/definition/application-release-automation-ARA">deploy their applications</a>. Automation helps businesses progress from commit and build to testing and deployment in a more systematic manner, improving efficiency and throughput while reducing human error. By using IT automation, businesses can deploy their applications with confidence, configure necessary services from the outset and retrieve their applications and artifacts.</li> <li><b>Security and compliance. </b>IT operations managers can use automation to define and enforce security, compliance and risk management policies and to remediate issues by building these as automated steps across their infrastructure. Automation enables IT operations managers to keep security at the front of their IT processes and to be more proactive in their security efforts. Incorporating standardized, automated cybersecurity processes and workflows makes compliance and auditing easier. Similarly, automation can be triggered by security events, initiating an immediate and consistent response to detected threats.</li> </ul> </section> <section class="section main-article-chapter" data-menu-title="Benefits of IT automation"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Benefits of IT automation</h2> <p>IT automation provides process efficiency and consistency, business agility, compliance, AI assistance and lower costs. More specifically, <a href="https://www.techtarget.com/searchitoperations/feature/Benefits-and-challenges-of-IT-automation">IT automation offers the following benefits</a>:</p> <figure class="main-article-image full-col" data-img-fullsize="https://www.techtarget.com/rms/onlineimages/it_automations_benefits_and_challenges-f.png"> <img data-src="https://www.techtarget.com/rms/onlineimages/it_automations_benefits_and_challenges-f_mobile.png" class="lazy" data-srcset="https://www.techtarget.com/rms/onlineimages/it_automations_benefits_and_challenges-f_mobile.png 960w,https://www.techtarget.com/rms/onlineimages/it_automations_benefits_and_challenges-f.png 1280w" alt="Graphic showing the pros and cons of IT automation." height="336" width="560"> <figcaption> <i class="icon pictures" data-icon="z"></i>Along with the benefits of IT automation come certain challenges. </figcaption> <div class="main-article-image-enlarge"> <i class="icon" data-icon="w"></i> </div> </figure> <ul type="disc" class="default-list"> <li>IT automation, real-time data and analytics empower companies to quickly adapt to the changing business environment, extend their reach to new locations and target different demographics.</li> <li>With the demands of today's distributed public, private and hybrid operating environments -- as well as the movement of data, applications and workloads among clouds -- IT automation systems enable businesses to handle higher workloads and data transactions without compromising performance.</li> <li>Automating tedious, repetitive, manual tasks and streamlining processes reduces time spent and costly errors by eliminating the human factor, improves consistency and accuracy, speeds up response times by IT teams as needs arise, and frees up workers to focus on more important, complex projects.</li> <li>IT automation platforms with embedded AI and machine learning (ML) capabilities can do more than automate repetitive rules-based tasks. With the advent of generative AI (GenAI) and agentic AI, these platforms can help solve problems, broaden the scope of <a href="https://www.techtarget.com/searchitoperations/definition/task-automation">automated tasks</a> and provide better support for developing technologies.</li> <li>Different IT administrators perform the same task in different ways, and even the same administrator handles a task differently from one time to the next. For corporate governance and regulatory compliance, an IT automation strategy helps ensure consistency in IT operations, regardless of which administrator is on duty on any given day.</li> <li>IT automation tools must be compatible with systems, software and other elements across potentially diverse IT environments. Integration with higher-level orchestration tools bundles tasks into governed workflows.</li> <li>The orchestration of automated tasks provides detailed reporting and visibility into the processes and workflows to identify anomalies, make necessary adjustments and ensure uninterrupted operations.</li> <li>Automated frameworks and ML tools can automate workflows and repetitive tasks used in systems management and network maintenance to help companies improve their security posture.</li> <li>Automating repeatable tasks can lower costs for infrastructure management, cloud services, application deployment, test environments and security incidents.</li> <li>Business users with fewer technical skills or without the assistance of IT professionals -- sometimes known as <i>citizen developers</i> -- can create web apps, mobile apps and workflows using low-code/no-code platforms and AI code assistants.</li> </ul> </section> <section class="section main-article-chapter" data-menu-title="Challenges of IT automation"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Challenges of IT automation</h2> <p>The benefits of IT automation don't always guarantee results. IT teams must be competent and skilled in using IT automation tools to translate behaviors into concrete procedural steps. If not, <a href="https://www.techtarget.com/searchitoperations/tip/IT-automation-challenges-and-how-to-overcome-them">IT automation can pose several challenges</a>, including the following:</p> <ul type="disc" class="default-list"> <li>Establishing an automated task requires a clear understanding of technical and business goals, which can take time, effort and expense to create, review, validate and update.</li> <li>Although automated tasks can be simple and straightforward, some tasks might be extremely complex and demand extensive expertise to implement properly -- especially for businesses with legacy systems on-premises, multiple platforms and large volumes of data.</li> <li>Businesses often need to invest in API management software and integration platforms because legacy automation tools might not support APIs for cloud services.</li> <li>Automation trades flexibility for speed and efficiency, making it challenging to update automation elements or respond quickly to changing business needs and unexpected events.</li> <li>IT automation systems can be prone to error and might require modifications -- and an automated error, by its nature, proliferates much more quickly than a manual error. Also, changing requirements, unstable interfaces and code updates can make maintaining and fixing IT automation more difficult than implementing it.</li> <li>Building an automation team can require staff with specialized skills in automation design, cloud architecture, infrastructure and capacity management, systems automation, software development, <a href="https://www.techtarget.com/searchitoperations/definition/security-automation">security automation</a>, testing frameworks and AI.</li> <li><a href="https://www.techtarget.com/searchitoperations/feature/IT-process-automation-6-examples-to-boost-efficiency">IT process automation</a> could change how some IT departments and other employees perform their jobs. These changes could require extensive collaboration and a willingness to develop new skills around AI usage and task optimization.</li> <li>Some businesses can fall into the trap of over-automation -- for example, automating complex tasks that require manual intervention or investing in technologies without the necessary skill sets on board to implement them.</li> <li>In their rush to invest in IT automation, businesses might overlook the steps to ensure the processes they're trying to automate will create a significant ROI.</li> <li>Automation elements are often dependent on underlying tools, platforms and frameworks, and, therefore, can result in vendor lock-in. Changing tool versions or adopting new tools could require reworking automation elements.</li> </ul> <figure class="main-article-image full-col" data-img-fullsize="https://www.techtarget.com/rms/onlineimages/top_areas_for_process_automation_efficiency_wins-f.png"> <img data-src="https://www.techtarget.com/rms/onlineimages/top_areas_for_process_automation_efficiency_wins-f_mobile.png" class="lazy" data-srcset="https://www.techtarget.com/rms/onlineimages/top_areas_for_process_automation_efficiency_wins-f_mobile.png 960w,https://www.techtarget.com/rms/onlineimages/top_areas_for_process_automation_efficiency_wins-f.png 1280w" alt="Graphic listing industry expert quotes on successful process automation applications. " height="280" width="560"> <figcaption> <i class="icon pictures" data-icon="z"></i>Industry experts point to successful process automation applications. </figcaption> <div class="main-article-image-enlarge"> <i class="icon" data-icon="w"></i> </div> </figure> </section> <section class="section main-article-chapter" data-menu-title="IT automation tools and vendors"> <h2 class="section-title"><i class="icon" data-icon="1"></i>IT automation tools and vendors</h2> <p><a href="https://www.techtarget.com/searchitoperations/tip/IT-process-automation-platforms-to-explore">IT automation platforms</a>, tools and frameworks automate tasks ranging from repetitive operations to <a href="https://www.techtarget.com/searchitoperations/tip/Open-source-workflow-engines-and-how-to-use-them">complex workflows</a>, including testing, configuration management and orchestration. IT automation software traditionally required some coding ability because many tools lacked low-code, drag-and-drop functionality to configure automated workflows. Now, many IT automation platforms support low-code, no-code or AI-assisted workflow creation.</p> <p>The sheer number of IT automation tools can complicate the selection process. Selecting the wrong tool can lead to deployment failures, workflow disruptions, higher costs and tool sprawl. It's important to understand the various categories of IT automation tools and their applications, with consideration geared toward business needs, system integration, scalability, compliance demands, degree of difficulty and performance monitoring. The following are common IT automation tool categories and their primary functions:</p> <ul type="disc" class="default-list"> <li><b>Provisioning and software deployment. </b>These tools<b> </b>automate tasks such as setting up user accounts, deploying OS images and installing or updating software on servers and PCs.</li> <li><b>Configuration management.</b> CM tools define, apply and maintain configurations as part of software development, deployment and change management.</li> <li><b>Workflow automation.</b> These<b> </b>platforms automate business and technical workflows involving multiple tasks and often serve as the foundation for orchestration.</li> <li><b>API management.</b> Tools in this category support API creation, publication, security and operation to control data and service access through a centralized gateway for developers.</li> <li><b>Continuous integration and continuous delivery. </b>CI/CD tools automate software building, testing and deployment throughout the software delivery pipeline.</li> <li><b>Monitoring and anomaly detection.</b> These types of platforms&nbsp;<a href="https://www.techtarget.com/searchitoperations/definition/continuous-monitoring">monitor applications</a> and infrastructure. They detect anomalies that might signal performance, application health, availability or security issues.</li> <li><b>AIOps and AI assistants.</b> These tools can automate or assist with tasks such as incident detection, alert analysis, root cause analysis, <a href="https://www.techtarget.com/searchitoperations/definition/What-is-an-automated-script">script generation</a>, infrastructure provisioning and configuration management.</li> <li><b>Task automation.</b> These platforms <a href="https://www.techtarget.com/searchitoperations/tip/Task-automation-tools-to-increase-productivity">automate many aspects of daily business operations</a>, including notifications, alerts, status updates and data movement and backup.</li> <li><b>Service desk automation and virtual assistants.</b> With these tools, IT teams can <a href="https://www.techtarget.com/searchitoperations/feature/service-desk-automation-examples-to-enhance-IT-support">streamline ticket management</a> and user support through chatbots, ChatOps tools and automated workflows.</li> <li><b>Governance and compliance.</b> Software in this category manages regulatory and legal issues, such as digital rights control, data security and protection, business continuity and disaster recovery, as well as legal records and documentation.</li> <li><b>Infrastructure as code.</b> IaC software uses configuration files or programming languages to provision and manage infrastructure through consistent, repeatable deployments across physical, virtual and cloud environments.</li> <li><b>Kubernetes platforms and GitOps.</b> These tools automate the deployment, configuration and management of containerized applications using declarative configurations maintained in version control.</li> </ul> <div class="youtube-iframe-container"> <iframe id="ytplayer-0" src="https://www.youtube.com/embed/xV-bOHvfbPQ?autoplay=0&amp;modestbranding=1&amp;rel=0&amp;widget_referrer=null&amp;enablejsapi=1&amp;origin=https://www.techtarget.com" type="text/html" height="360" width="640" frameborder="0"></iframe> </div> <p>Many vendors operate in the IT automation landscape. Some are dedicated to specific automation tools, while others are major technology providers that offer automation as part of their wide-ranging product offerings.</p> <p>Microsoft's automation offerings include System Center Orchestrator for IT infrastructure automation, Service Manager for IT service management, PowerShell for scripting and system configuration and PowerShell Desired State Configuration for automating Windows and Linux configurations. Microsoft also offers Azure Automation for cloud automation.</p> <p>Broadcom and BMC provide enterprise automation tools for provisioning, patching, configuration and compliance management across physical, virtual and cloud environments, with BMC offering preconfigured compliance policies.</p> <p>Red Hat Ansible, Pulumi, Chef, Puppet, Salt and HashiCorp Terraform support automation and IaC, helping organizations create consistent workflows from development through IT operations.</p> <p>IT professionals who prefer general-purpose workflow automation can use business process automation platforms such as Zapier, ClickUp, Kissflow and ProcessMaker to automate processes, including ticketing, incident management, user provisioning, asset tracking and reporting.</p> </section> <section class="section main-article-chapter" data-menu-title="How to build a business case for automation"> <h2 class="section-title"><i class="icon" data-icon="1"></i>How to build a business case for automation</h2> <p>IT has evolved from a business burden to a vital service that must support and keep pace with constant change. Traditional, manual IT management can't meet today's business needs, and technologies such as automation and orchestration have become indispensable for modernizing business processes.</p> <p>IT teams must do their homework and apply proof-of-concept principles to process and workflow modernization, which -- when done properly -- present the following advantages:</p> <ul type="disc" class="default-list"> <li>Automation can deliver consistently successful outcomes without errors or oversights commonly associated with manual human intervention.</li> <li>Automation can handle greater workloads faster than manual human effort.</li> <li>Greater speed and less human intervention can reduce the business costs associated with many important tasks and workflows.</li> <li>With relief from the burden of time-consuming and repetitive tasks, professionals can focus on more strategic actions that are more valuable to the business.</li> <li>Automation's consistency helps organizations implement proper security checks and regulatory compliance requirements.</li> </ul> <p>The business case should not only focus on demonstrating the overall benefits of automation, but it should also zero in on the key areas most in need of automation and easiest to automate and measure, to ensure early successes. A sound business case depends on an ongoing comprehensive strategy that encompasses every aspect of IT automation, including business goals, governance practices, tool selection and proper training.</p> <figure class="main-article-image full-col" data-img-fullsize="https://www.techtarget.com/rms/onlineimages/building_an_it_automation_strategy-f.png"> <img data-src="https://www.techtarget.com/rms/onlineimages/building_an_it_automation_strategy-f_mobile.png" class="lazy" data-srcset="https://www.techtarget.com/rms/onlineimages/building_an_it_automation_strategy-f_mobile.png 960w,https://www.techtarget.com/rms/onlineimages/building_an_it_automation_strategy-f.png 1280w" alt="Graphic showing 10 steps in building an IT automation strategy." height="392" width="560"> <figcaption> <i class="icon pictures" data-icon="z"></i>There's much to consider before and after deploying IT automation. </figcaption> <div class="main-article-image-enlarge"> <i class="icon" data-icon="w"></i> </div> </figure> <p>A well-designed strategy should detail and document the approach an enterprise will use to automate tasks, speed up workflows across the infrastructure, reduce errors and delays caused by human intervention, and deliver necessary IT services faster and at lower cost than manual processes.</p> </section> <section class="section main-article-chapter" data-menu-title="Best practices for implementing IT automation"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Best practices for implementing IT automation</h2> <p>The <a href="https://www.techtarget.com/searchitoperations/tip/Follow-these-8-steps-to-implement-automation-in-IT-workflows">path to implementing automation</a> is fraught with mistakes and waste. Careful planning and a concerted effort are critical to implementing automation plans in a way that's meaningful and maintainable. Best practices for implementing IT automation include the following:</p> <ul type="disc" class="default-list"> <li>Establish clear goals for the automation initiative, including problems the business is trying to solve and how automation tools and tactics will address those issues now and five years after implementation.</li> <li>Know that automation isn't all or nothing. Target common and frequently performed tasks that consume the most resources, demand significant time and are most error-prone.</li> <li>In addition to targeting a certain task for automation, consider the current workflows for multiple tasks companywide and seek common sequences or subprocesses for automation that can lead to orchestration.</li> <li>All tasks and workflows identified for automation should be carefully reviewed and approved by the various stakeholders, including IT, business, security and legal teams, to eliminate traditional silos of responsibility that can sabotage implementation.</li> <li>Evaluate IT automation tools, platforms and frameworks that best suit current and future automation projects. They can range from IT-centric tools to business workflow automation platforms.</li> <li>Implement a series of test projects so participants can learn about the tools and develop a working knowledge of automation development and implementation.</li> <li>Integrate IT automation tools with change management systems to ensure changes are reviewed, appropriate, implemented correctly by responsible teams and audited if necessary.</li> <li>Apply metrics or KPIs to objectively measure how automation affects the business. Monitoring also helps determine bottlenecks within the automated processes.</li> <li>Automate a small suite of tasks and workflows and prove automation's value to the business, then systematically automate additional tasks.</li> <li>Plan periodic reviews to revisit and revalidate automated processes, so automation maintenance is not neglected or treated as an afterthought.</li> </ul> <figure class="main-article-image full-col" data-img-fullsize="https://www.techtarget.com/rms/onlineimages/12_steps_to_automate_it_tasks-f.png"> <img data-src="https://www.techtarget.com/rms/onlineimages/12_steps_to_automate_it_tasks-f_mobile.png" class="lazy" data-srcset="https://www.techtarget.com/rms/onlineimages/12_steps_to_automate_it_tasks-f_mobile.png 960w,https://www.techtarget.com/rms/onlineimages/12_steps_to_automate_it_tasks-f.png 1280w" alt="Graphic showing the 12 steps to automating IT tasks." height="336" width="560"> <figcaption> <i class="icon pictures" data-icon="z"></i>Automating IT tasks includes workflow evaluation, security considerations, tool selection and performance monitoring. </figcaption> <div class="main-article-image-enlarge"> <i class="icon" data-icon="w"></i> </div> </figure> </section> <section class="section main-article-chapter" data-menu-title="IT automation teams: Roles, skills and cultural needs"> <h2 class="section-title"><i class="icon" data-icon="1"></i>IT automation teams: Roles, skills and cultural needs</h2> <p>Nowhere within an enterprise is collaboration more important than IT automation, which powers everything from application development and infrastructure deployment to business processes. The core responsibilities of IT automation teams center on integrating systems, applications and data; creating scripts and APIs to connect systems; and automating tasks and workflows.</p> <p>A typical IT automation team consists of stakeholders from practically every corner of the enterprise. They can include IT professionals, system <a href="https://www.techtarget.com/searchitoperations/definition/automation-architect">automation specialists</a>, cloud <a href="https://www.techtarget.com/searchitoperations/definition/automation-engineer">automation engineers</a>, DevOps engineers, AI developers, security automation specialists, site reliability engineers, Agile coaches, integration engineers and business process engineers.</p> <p>These roles typically require experience in some of the following practices:</p> <ul class="default-list"> <li>Agile and <a href="https://www.techtarget.com/searchitoperations/feature/AIOps-vs-DevOps-Distinct-approaches-to-IT-automation">DevOps principles</a>.</li> <li>AI and ML.</li> <li>Automation design.</li> <li>Cloud architecture.</li> <li>Infrastructure and capacity management.</li> <li>Security automation.</li> <li>Software development.</li> <li>Systems automation.</li> <li>Testing frameworks.</li> </ul> <p>Fulfilling these roles could entail a wide range of technical skills and knowledge, such as scripting (Python, JavaScript, Bash and PowerShell), source code management; containers and <a href="https://www.techtarget.com/searchitoperations/tip/Kubernetes-automation-Use-cases-and-tools-to-know">Kubernetes</a>; security; testing; observability; monitoring; networks; and business and industry. Additional important soft or cultural skills include leadership, problem-solving, collaboration, communication, storytelling and a focus on automation.</p> <p>All these roles and responsibilities work together to ensure a continuous cycle of improvement in IT automation practices.</p> </section> <section class="section main-article-chapter" data-menu-title="Trends and future of IT automation"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Trends and future of IT automation</h2> <p>As automation tools become faster, easier and cheaper to deploy, organizations are expanding their use across more processes and workflows. The common thread among <a href="https://www.techtarget.com/searchitoperations/feature/6-IT-automation-trends-to-watch-in-2025-and-the-future">IT automation trends</a> is the convergence of technologies and the orchestration of end-to-end processes as businesses shift toward more intelligent, efficient and interconnected operations.</p> <p>AI and ML add a <a href="https://www.techtarget.com/searchenterpriseai/definition/intelligent-process-automation-IPA">layer of intelligence</a> that transforms IT automation from simple, task-oriented machine applications into systems that can learn from data, identify patterns, make predictions, support and automate decision-making, bring together disparate tasks and processes, and handle more complex work without human intervention.</p> <h3>Agentic AI and intelligent automation</h3> <p>Agentic AI adoption is increasing. Intelligent AI agents are moving a step closer to humanlike behavior with the capacity to understand their surroundings, reason things out and immediately act on their own -- with guardrails -- which could speed up productivity.&nbsp;Yet many businesses planning to increase their automation investments are struggling to integrate agentic AI and AI agents into their processes.</p> <p>Hyperautomation, which combines AI, ML and robotic process automation (RPA), is also experiencing a resurgence. GenAI-infused tools can help ensure smoother, more cohesive end-to-end processes by working with legacy systems and gluing together once-isolated parts of automation. RPA is shedding its role as a one-dimensional rules-based system limited to simple, repetitive tasks and instead taking on greater intelligence by integrating with AI and learning to adapt to its environment.</p> <figure class="main-article-image full-col" data-img-fullsize="https://www.techtarget.com/rms/onlineimages/keep_an_eye_on_these_it_automation_trends-f.png"> <img data-src="https://www.techtarget.com/rms/onlineimages/keep_an_eye_on_these_it_automation_trends-f_mobile.png" class="lazy" data-srcset="https://www.techtarget.com/rms/onlineimages/keep_an_eye_on_these_it_automation_trends-f_mobile.png 960w,https://www.techtarget.com/rms/onlineimages/keep_an_eye_on_these_it_automation_trends-f.png 1280w" alt="Graphic showing the top trends in IT automation." height="301" width="559"> <figcaption> <i class="icon pictures" data-icon="z"></i>IT automation trends emphasize GenAI, agentic AI and convergence. </figcaption> <div class="main-article-image-enlarge"> <i class="icon" data-icon="w"></i> </div> </figure> <h3>Orchestration and automation platforms</h3> <p>Business orchestration and automation technologies will converge into platforms that include aspects of RPA, digital process automation, integration platform as a service and low-code tools to take on a wider swath of business processes. These platforms will focus on orchestration, agent-building using prompts and new forms of agent governance.</p> <p>AI-driven orchestration is also expanding, enabling systems to execute tasks, predict problems and proactively optimize workflows with greater autonomy, accuracy and speed than manual implementations. This enables IT teams to accomplish far more work in the same amount of time. AI also provides detailed reporting, visibility and analysis, so orchestrated workflows can deliver predictable and repeatable outcomes that help companies demonstrate their preparedness for security, compliance and business continuity.</p> <h3>Expanding access to automation</h3> <p>Also on the rise is no-code/low-code automation, which enables users to create their own workflows without programming expertise. Self-service automation will enable business users to build their own automations using RPA software and eventually incorporate GenAI components. Citizen developers are expected to build a significant portion of <a href="https://www.techtarget.com/searchitoperations/tip/Beyond-automation-Using-GenAI-to-modernize-IT-operations">GenAI-infused automation</a> applications, primarily developing initial workflows, creating forms and visualizing processes.</p> <p>AI-assisted automation tools in software development are continuously evolving. CI/CD workflow tools are well established, but future activity will emphasize automation at the code-creation level.</p> <h3>Extending automation across IT operations</h3> <p>Security automation, which detects and responds to incidents, is becoming more widespread. Cybersecurity automation tools will need to be adaptable and insightful so&nbsp;AI can build context about network activity&nbsp;and react to changing business goals and needs.</p> <p>Another trend, edge computing, brings data closer to the source for real-time analysis. Business ecosystem automation is an overarching concept of using technology to streamline and automate processes within a network of interconnected businesses, partners and customers.</p> <h3>Emerging AI capabilities and foundations</h3> <p>Embodied AI, aided by edge intelligence, <a target="_blank" href="https://www.sciencedirect.com/science/article/pii/S209580992500815X" rel="noopener">is being integrated</a> into robots, enabling them to interact with and learn from their environment through sensors, motors and ML. Instead of following preprogrammed rules and workflows, these robots can sense and respond to their environment to handle more complex and unpredictable situations -- much like agentic AI systems are anticipated to do in automated processes.</p> <p>Internal LLMs with proprietary data and retrieval-augmented generation (RAG) capabilities hold even more promise for IT automation. RAG enables AI to retrieve relevant information while it's generating responses. AI-driven automation depends on trusted, accurate data and effective governance, as poor-quality data, biased models and security risks can undermine automation initiatives.</p> <p><b>Editor's note:</b><i> This article was updated in 2026 to reflect the latest IT automation advancements and applications.</i></p> <p><i>Stephen J. Bigelow, senior technology editor at Informa TechTarget, has more than 30 years of technical writing experience in the PC and technology industry.</i></p> <p><i>Ron Karjian is an industry editor and writer at Informa TechTarget covering business analytics, artificial intelligence, data management, security and enterprise applications.</i></p> <p><i>Kinza Yasar and Ryann Burnett also contributed to this article.</i></p> </section> IT automation uses instructions to create clear, consistent and repeatable processes that replace an IT professional's manual work in data centers and cloud deployments. https://cdn.ttgtmedia.com/visuals/digdeeper/6.jpg https://www.techtarget.com/it-infrastructure/definition/What-is-IT-automation-A-complete-guide-for-IT-teams Tue, 01 Sep 2026 10:17:00 GMT <p>Vector database specialist Qdrant on Tuesday launched Fineweb-10B, a public dataset for enterprises to use when benchmarking their vector retrieval capabilities.</p> <p>The dataset uses information from FineWeb, an open-source dataset created by <a href="https://www.techtarget.com/whatis/definition/Hugging-Face">Hugging Face</a> that contains more than 15 trillion pieces of text for training large language models.</p> <p>Vector embeddings have become <a href="https://www.techtarget.com/data-technologies/feature/Vector-search-now-a-critical-component-of-GenAI-development">a critical part of the AI pipeline</a>. Vectors are algorithmically assigned numerical representations that make data -- including <a href="https://www.techtarget.com/data-technologies/post/Unstructured-data-is-the-bottleneck-for-agentic-AI-success">unstructured data</a> -- easy to search and discover. Once discovered, whether by agents or human users, vectorized data can be used to inform analytics and AI tools.</p> <p>However, many organizations don't fully understand the speed and accuracy of their vector retrieval systems at the scale now demanded by AI development, and whether their vectorized data can be trusted to inform applications in production environments. As a result, Qdrant's launch of a massive public dataset for testing vector retrieval speed and accuracy at enterprise scale is valuable, according to William McKnight, president of McKnight Consulting.</p> <p>"Having done a number of benchmarks evaluating [large-scale retrieval], I can attest there is a&nbsp;massive compute barrier needed to calculate exact ground truth across 10 billion vectors," he told TechTarget. "This dataset&nbsp;enables organizations to accurately measure recall … on realistic, non-synthetic web data under true production scale."</p> <p>In addition, building the dataset with capabilities from Hugging Face, the <a target="_blank" href="https://commoncrawl.org/" rel="noopener">Common Crawl</a> Foundation, Vultr and Alibaba lends it credibility, McKnight continued.</p> <p>"Because enterprise IT teams lack the dedicated methodology and neutral vantage point to properly configure and stress-test these massive workloads and vendors lack independence, partnering with an independent, reputable third party is essential to obtain credible, production-grade evaluations free from vendor bias," he said.</p> <section class="section main-article-chapter" data-menu-title="Open-source value"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Open-source value</h2> <p>Qdrant's origins are in open-source technology.</p> <p>Founded in 2021 and based in Berlin and New York City, the vendor began as an open-source project published on GitHub before launching a proprietary version of its vector database in 2023. Meanwhile, with both versions of its vector database built with the open-source <a href="https://rust-lang.org/">Rust programming language</a>, Qdrant remains closely tied to the open-source community.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> This dataset enables organizations to accurately measure recall … on realistic, non-synthetic web data under true production scale. </figure> <figcaption> <strong>William McKnight</strong>President, McKnight Consulting </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>Developing a public dataset for benchmarking vector retrieval capabilities reflects the vendor's recognition that public benchmarks that measure the rising demands being placed on vector retrieval systems are needed, according to Nathan LeRoy, a research engineer at Qdrant Labs.</p> <p>"Vector search workloads have already hit billion vector scale. Enterprises need benchmarks to reflect this so that they can make informed decisions at this scale," he told TechTarget. "Ideally, such benchmarks and their datasets are public and accessible, so it's easy to make comparisons and draw conclusions across search systems."</p> <p>While Qdrant recognizes the value of the open-source tools, McKnight noted that vendors sometimes release open-source capabilities even if they aren't closely aligned with the open-source community.</p> <p>Google, for example, developed the Agent2Agent protocol in May 2025 to provide an open standard for connecting and <a href="https://www.techtarget.com/enterprise-software/news/366636690/Agentic-orchestration-the-next-AI-issue-for-CIOs-to-tackle">orchestrating agents</a> and donated it to the Linux Foundation.</p> <p>However, the reasons vendors develop open-source capabilities are not always magnanimous, McKnight continued. While it shows that the vendor is committed to its market and is aligned with cutting-edge thought, it also can lead to greater visibility and attract new potential customers.</p> <p>"It isn't uncommon for vendors to release open-source benchmarking datasets and tools, especially ones they are likely good at competitively," he said.</p> <p>Michael Ni, an analyst at Constellation Research, similarly noted that Qdrant's launch of an open-source dataset for benchmarking vector retrieval is beneficial to enterprises as they <a href="https://www.techtarget.com/ai/tip/Citizen-developers-are-redefining-enterprise-AI-development">build AI tools,</a> as well as the vendor itself.</p> <p>"A public dataset like Fineweb-10B is an ecosystem and category-building move," he told TechTarget. "Qdrant is effectively subsidizing the cost of evaluation … and position[ing] Qdrant as central to that evaluation. Qdrant gains mindshare, creates de facto standards and evaluation criteria, and increases the odds [potential customers] ask the kinds of questions Qdrant is already optimized to answer."</p> </section> <section class="section main-article-chapter" data-menu-title="Up to the test"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Up to the test</h2> <p>Qdrant Fineweb-10B is aimed at enabling organizations to benchmark vector retrieval at enterprise scale.</p> <p>Toward that end, Qdrant built the dataset using vector embeddings of real documents rather than <a href="https://www.techtarget.com/searchenterpriseai/tip/What-to-know-about-synthetic-data-as-a-business-advantage">synthetic data</a>, and a corpus -- a collection of data -- of 10 billion documents so it is large enough to ensure that size-dependent behavior shows up during testing. In addition, the dataset includes a grounded truth query set, which is a list of search prompts paired with their ideal responses against which an enterprise can measure the accuracy of its benchmark tests.</p> <p>Ni noted that Fineweb-10B is not the first public dataset for benchmarking vector retrieval. However, the size of the new dataset differentiates it from others.</p> <p>"This is an important extension of existing benchmark work, not a brand-new category of benchmarking," he said. "Qdrant has contributed a much larger, more application-like retrieval workload with known correct answers that allow engineering teams' tests to go beyond throughput to how accurately the system behaves as corpus size, filtering and retrieval complexity increase."</p> <p>Meanwhile, the dataset is appropriately constructed to aid organizations seeking to better understand the effectiveness of their vector retrieval systems, Ni continued, while adding that it is only one part of what should be <a href="https://www.techtarget.com/it-infrastructure/tip/How-effective-is-your-AI-agent-9-benchmarks-to-consider">a broader evaluation process</a>.</p> <p>"Qdrant-Fineweb-10B certainly solves&nbsp;the problem Qdrant says it is solving, measuring whether vector retrieval&nbsp;remains accurate and performant at very large scale," Ni said. "At the same time, companies still need to test against their own data, business filters, re-ranking and downstream agent outcomes as part of a total enterprise retrieval approach versus just testing a retrieval engine."</p> <p>McKnight similarly noted that Qdrant-Fineweb-10B is well constructed, though limited in scope.</p> <p>"The dataset seems well-designed for pairing realistic web-text distributions with exact ground truth across dense, sparse and filtered queries," he said. "However, fully leveraging it seems to require substantial compute and storage infrastructure to ingest and index 10 billion records, and it does not test real-time streaming updates."</p> <figure class="main-article-image full-col" data-img-fullsize="https://www.techtarget.com/rms/onlineimages/traditional_search_vs_vector_search-f.png"> <img data-src="https://www.techtarget.com/rms/onlineimages/traditional_search_vs_vector_search-f_mobile.png" class="lazy" data-srcset="https://www.techtarget.com/rms/onlineimages/traditional_search_vs_vector_search-f_mobile.png 960w,https://www.techtarget.com/rms/onlineimages/traditional_search_vs_vector_search-f.png 1280w" alt="A graphic shows the differences between keyword search and vector search. " data-credit="Informa TechTarget " height="319" width="560"> <div class="main-article-image-enlarge"> <i class="icon" data-icon="w"></i> </div> </figure> </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Looking ahead</h2> <p>As Qdrant plans product development, speed, accuracy and openness are focal points, according to LeRoy. In April, the vendor's <a href="https://www.techtarget.com/data-technologies/news/366642580/Qdrant-boosts-performance-reliability-to-meet-AI-needs">Qdrant Cloud update</a> included features aimed at improving performance.</p> <p>"As 2026 draws to a close, we’ll continue to push the limits of vector search for speed, accuracy, and being deployable anywhere," LeRoy said.</p> <p>Focusing on speed and accuracy is wise, according to Ni.</p> <p>He noted that vector database specialists such as Qdrant and Pinecone now face competition from broad-based data management providers, including hyperscale cloud vendors <a href="https://www.techtarget.com/data-technologies/news/366577632/Vector-search-and-storage-key-to-AWS-database-strategy">AWS</a> and <a href="https://www.techtarget.com/data-technologies/news/366583139/Oracle-adds-vector-search-capabilities-to-database-platform">Oracle</a>, that provide vector search. As a result, specialists need to distinguish themselves.</p> <p>"Vector search is becoming table stakes," Ni said. "Most enterprises can get it from databases they already own, so specialists like Qdrant have to drive what makes them distinct and valuable to the market by delivering better retrieval quality, latency and cost at scale."</p> <p>McKnight, meanwhile, suggested that better integrating <a href="https://www.techtarget.com/searchitchannel/definition/feedback-loop">feedback loops</a> and adding to its <a href="https://www.techtarget.com/data-technologies/tip/Top-5-metadata-management-best-practices">metadata management</a> capabilities would help Qdrant better serve existing customers and perhaps draw in new ones.</p> <p>"By natively integrating generative feedback loops, [Qdrant would remove] the need for teams to build these pipelines via external APIs," he said. "Additionally, broadening its rich scalar metadata capabilities to include multidimensional matrices and tensors would enhance support for complex data structures."</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> With AI tools requiring massive volumes of vectors to perform properly, the public dataset is designed to help organizations test their systems in real-world conditions. https://cdn.ttgtmedia.com/rms/onlineimages/code_g1195673150.jpg https://www.techtarget.com/data-technologies/news/366649627/Qdrant-builds-dataset-to-benchmark-vector-retrieval-at-scale Tue, 01 Sep 2026 09:00:00 GMT <p>As AI creates challenging conditions for analytics specialists by becoming the new interface for BI, Tableau is evolving.</p> <p>And it's <a href="https://www.techtarget.com/data-technologies/feature/New-Tableau-leader-talks-vendors-evolution-in-era-of-AI">doing so</a> in an appropriate manner, according to David Menninger, an analyst at ISG Software Research.</p> <p>"Tableau continues to be a standard bearer," he told TechTarget. "They are not alone, but they eat, sleep and breathe analytics, and are constantly pushing the boundaries."</p> <p>AI is altering the way users consume BI, making some of the chief capabilities that Tableau and competitors such as Qlik and Microsoft excelled at providing obsolete. To remain relevant, <a href="https://www.techtarget.com/data-technologies/news/366645161/Tableau-Qlik-in-flux-What-it-means-for-BI-users-amid-AI-shift">such vendors</a> need to adjust.</p> <p>Tableau, a subsidiary of CRM giant Salesforce specializing in analytics, hosted a webinar on Aug. 24 during which it previewed the features it plans to highlight during Dreamforce, Salesforce's annual user conference which will be held Sept. 15-17 in San Francisco.</p> <p>None are new capabilities, though each has features that are not yet generally available. They include <a href="https://www.techtarget.com/data-technologies/news/366642778/Tableau-repositions-for-AI-unveils-new-knowledge-layer">the knowledge layer</a> that Tableau unveiled in May as part of its new Agentic Analytics Platform, <a href="https://www.techtarget.com/it-infrastructure/tip/An-overview-of-headless-architecture-design">headless BI capabilities</a> that enable customers to analyze data while using applications such as Slack and ChatGPT, and proactive intelligence that surfaces insights that might not otherwise have been discovered.</p> <p>Collectively, the capabilities demonstrate that Tableau is changing in a logical way, according to Donald Farmer, founder and principal of TreeHive Strategy. But he suggested that there is more the vendor should do to keep current, and he noted that even if Tableau does make all the right moves as it reacts to a new reality, AI's emergence drastically alters the paradigm for traditional BI providers.</p> <p>"Tableau is right to focus on context, the one thing AI models do not supply," Farmer told TechTarget. "Salesforce owns as much of the enterprise context -- within its domain -- as anyone, so it's a strong position. Meanwhile, headless delivery is unavoidable and the right direction."</p> <p>However, while Tableau is correct to provide headless BI capabilities, doing so makes the vendor's platform less visible <a href="https://www.techtarget.com/data-technologies/feature/Telecom-giant-uses-Tableau-Pulse-to-drive-analytics-adoption">to its customers</a> by giving them a reason not to open their Tableau instance, he continued. In addition, simply providing new capabilities is likely not enough for BI providers to retain users.</p> <p>"I am not sure [they] address what the 'DataFam' needs -- training, tooling and credentials for maintaining knowledge with the depth needed for AI, not just re-packaging BI insights," Farmer said.</p> <section class="section main-article-chapter" data-menu-title="An altered landscape"> <h2 class="section-title"><i class="icon" data-icon="1"></i>An altered landscape</h2> <p>Reports, dashboards and data visualizations used to be the end point for analytics. They were how business users and expert analysts consumed data and derived insights that informed strategic decisions. Tableau excelled at providing customers with the capabilities to build such tools. So did Qlik, Strategy, ThoughtSpot, Domo and Microsoft, among others.</p> <p>AI has <a href="https://www.techtarget.com/data-technologies/news/366649201/What-AI-native-analytics-startups-offer-that-BI-vendors-dont">changed that</a>.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Tableau is right to focus on context, the one thing AI models do not supply. Salesforce owns as much of the enterprise context -- within its domain -- as anyone, so it's a strong position. Meanwhile, headless delivery is unavoidable and the right direction. </figure> <figcaption> <strong>Donald Farmer</strong>Founder and principal, TreeHive Strategy </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>Now, rather than going into a BI environment to develop data products and analyze data, users can simply ask a question using natural language and their chatbot or agent can deliver a response in seconds that includes detailed analysis, along with a supporting report or visualization. In addition, unlike a static report or dashboard that can take hours or weeks to update, users can immediately ask follow-up questions of AI tools to get deeper insights.</p> <p>"Before AI, visualizations were the holy grail of BI," Menninger said. "Now, BI is all about conversational interfaces -- ask a question, see some visualizations, ask a follow up question, ask more questions and provide some instructions to prepare a report or presentation. These things happened before, but not in the software."</p> <p>Farmer likewise noted that AI has altered BI by making data artifacts a byproduct of <a href="https://www.techtarget.com/data-technologies/tip/Business-intelligence-challenges-intensify-as-AI-use-grows">AI-powered analysis</a> rather than the deliverable while insights have become the product.</p> <p>"The report, dashboard, or visualization used to be the deliverable and decisions followed from that," he said. "Now, and especially with agentic AI, the decision is the deliverable, and the artifact is a by-product of the process. In fact, a lot of analysis now happens with no BI tool involved at all. Someone shares a spreadsheet with an AI assistant and has a chart in seconds."</p> <p>Therein lies the ongoing challenge for the vendors that specialized in BI and whose platforms empowered users to make data-driven decisions, Farmer continued.</p> <p>While Tableau and its peers now provide chatbots and agents that drastically <a href="https://www.techtarget.com/data-technologies/news/365530077/BI-adoption-poised-to-break-through-barrier-finally">simplify insight-generation</a> and enable any user with proper authorization to query and analyze data, so do broader-based data platform providers that previously didn't directly compete with BI vendors.</p> <p>"Some of the problems BI vendors specialized in are now pretty much solved," Farmer said. "Most BI vendors are still trying to sell the interface, but that has disappeared into more-or-less universal chat experience or agent framework."</p> <p>Beyond data platform vendors such as Databricks and Snowflake, analytics specialists now face further competition from major application providers such as ERP software vendor <a href="https://www.computerweekly.com/blog/CW-Developer-Network/Workday-expands-Illuminate-introduces-Workday-Data-Cloud">Workday,</a> as well as AI vendors including <a href="https://www.techtarget.com/healthtechanalytics/news/366642098/OpenAI-launches-ChatGPT-for-Clinicians">OpenAI</a> that provide chat interfaces, according to Menninger.</p> <p>"AI has significantly reduced the barriers to entry in the analytics market … so the biggest challenge is a barrage of new competition," he said.</p> <p>In response, Tableau is changing.</p> </section> <section class="section main-article-chapter" data-menu-title="A new emphasis"> <h2 class="section-title"><i class="icon" data-icon="1"></i>A new emphasis</h2> <p>Once renowned for its pioneering data visualization capabilities, Tableau is now touting tools that aid customers building AI tools and others that use AI to deliver insights to users without forcing them to use Tableau's environment or even do the analysis on their own.</p> <p>In May, the vendor unveiled a knowledge layer as part of its new Agentic Analytics Platform.</p> <p>The knowledge layer is designed to be a data foundation that agents and other AI applications can tap into to <a href="https://www.techtarget.com/data-technologies/news/366646252/Context-is-king-as-agents-evolve-but-problems-persist">access the contextually relevant</a> information that enables them to carry out their prescribed tasks. Included is a data knowledge engine based on Tableau's decades of <a href="https://www.techtarget.com/data-technologies/feature/Why-enterprise-AI-depends-on-the-semantic-layer">semantic modeling</a>, a natural language interface, and built-in security and governance.</p> <p>Given that AI outputs are often misleading because they are not accessing the appropriate data, Tableau's knowledge layer is built to improve the accuracy of AI outputs so that agents can be trusted in production environments, according to Emma Annand, Salesforce's director or product marketing for agentic analytics.</p> <p>"Data leaders are finding that the quality of data is going to be what [enables] successful agentic analytics," she said during the webinar. "We have to think of data as a valuable resource, and knowledge is the unlock."</p> <p>Tableau is not alone among analytics specialists in developing a knowledge layer.</p> <p>In addition to Tableau, vendors such as <a href="https://www.techtarget.com/data-technologies/news/366626134/ThoughtSpot-evolving-as-BI-becomes-driven-by-AI">ThoughtSpot</a> and <a href="https://www.techtarget.com/data-technologies/news/366640059/GoodDatas-Context-Management-aims-to-make-AI-trustworthy">GoodData</a> have also pivoted to transform their expertise in data analysis into fuel for AI. Similarly, data management providers including Alteryx, AWS, Databricks, Goole Cloud, Informatica, Microsoft, Snowflake and Teradata have all introduced tools aimed at connecting agents with context.</p> <p>Nevertheless, Menninger noted that it is important for Tableau's futureto provide such capabilities, as dedicated BI environments disappear and AI becomes the interface for generating insights.</p> <p>"As an industry we need to solve the problem of context and semantics," he said. "We still don't know all the information that an enterprise collects and what it means. That knowledge is necessary to correctly interpret the data and perform the appropriate actions."</p> <p>Farmer likewise noted that, as enterprises continue to <a target="_blank" href="https://www.gartner.com/en/newsroom/press-releases/2026-1-15-gartner-says-worldwide-ai-spending-will-total-2-point-5-trillion-dollars-in-2026" rel="noopener">invest in AI development</a> and strive to <a target="_blank" href="https://www.deloitte.com/content/dam/assets-zone3/us/en/docs/services/consulting/2026/state-of-ai-2026.pdf" rel="noopener">move pilots into production</a>, building a knowledge layer is significant and could help Tableau stand out from competitors that aren't evolving away from traditional BI.</p> <p>"[Rather than] just repackaging BI insights, the knowledge layer is a differentiator," he said.</p> <p>Adding headless BI and proactive intelligence capabilities are also wise ways for Tableau to evolve and remain relevant to its users, according to Menninger.</p> <p>Historically, Tableau was its own environment. Through APIs and software development kits, headless analytics enables users to export Tableau into agents, collaboration platforms such as Slack and Teams, AI tools including Claude and ChatGPT, and applications where they do much of their work.</p> <p>Headless BI is not a new concept. For example, GoodData <a href="https://www.techtarget.com/data-technologies/news/252501721/GoodData-launches-overhauled-analytics-platform">launched headless analytics capabilities</a> in 2021. But new or not, Menninger noted that headless BI is a valuable addition for Tableau users.</p> <p>"Salesforce has made a big bet on headless," he said. "We had a previous round of headless in analytics and it didn't go over very well. … Conceptually though, it is the right thing to do. Analytics should never have been a separate discipline. It needs to be integrated with the core business processes it supports."</p> <p>Proactive intelligence, meanwhile, is AI-driven analysis that <a href="https://www.techtarget.com/data-technologies/feature/How-to-get-reliable-BI-insights-from-AI-augmented-analytics">surfaces insights</a> so users don't have to constantly comb data to discover new ideas.</p> </section> <section class="section main-article-chapter" data-menu-title="Competitive standing"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Competitive standing</h2> <p>While adding and emphasizing capabilities that are useful as AI makes traditional BI less necessary, Tableau is largely just keeping pace with <a href="https://www.techtarget.com/data-technologies/tip/8-top-self-service-analytics-tools">its competitors</a> rather than pacing the market, according to Menninger.</p> <p>However, even though other vendors are providing knowledge layer and headless BI capabilities, quality is way that Tableau can continue to distinguish itself, he continued.</p> <p>"Tableau continues to be a standard bearer," Menninger said. "They are not alone, but they eat, sleep and breathe analytics, and are constantly pushing the boundaries."</p> <p>Farmer likewise noted that Tableau is in step with its peers rather than launching market-moving new capabilities.</p> <p>For example, Qlik has made <a href="https://www.techtarget.com/data-technologies/news/366587232/Trusted-data-key-for-Qlik-as-it-develops-foundation-for-AI">the data foundation</a> its focus rather than the analysis layer while ThoughtSpot, which built its platform around AI-powered search from the time it first launched, continues to make analytics part of an AI workflow rather than a set of standalone data products.</p> <p>However, whether Tableau and other BI specialists can further evolve to remain critical to their customers is unclear, Farmer continued.</p> <p>"Every BI vendor is moving in the same direction, and it's mostly downhill from here," he said.</p> <p>One way that Tableau could potentially grow is by breaking its knowledge layer out from the Agentic Analytics Platform, Farmer suggested.</p> <p>By doing so, Tableau would enable customers to use their Tableau semantic models to ground any agent, whether built using Salesforce's development tools or not, while their <a href="https://www.techtarget.com/data-technologies/feature/Data-and-AI-governance-must-team-up-for-AI-to-succeed">governance</a> and audit trails remain in Tableau.</p> <p>"That would turn … risk into revenue, where Tableau gets paid when its knowledge and context is used somewhere else, instead of losing a seat," Farmer said.</p> <p>Menninger, meanwhile, noted that Tableau could serve its existing users and perhaps even attract new ones by adding more <a href="https://www.techtarget.com/data-technologies/definition/What-is-decision-intelligence">decision intelligence capabilities</a> such as <a href="https://www.techtarget.com/data-technologies/news/252523947/Scenario-planning-fertile-ground-for-analytics-vendors">scenario planning</a>.</p> <p>"Tableau, and other analytics providers, need to learn to live in an agentic world," he said. "As agents evolve, they will incorporate more and more decision making, [so] I expect to see more decision intelligence in analytics products. To make informed decisions, you need to evaluate multiple scenarios and alternatives. We still don't see enough of these capabilities in most BI products."</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> A knowledge layer and headless analytics capabilities show that the vendor is changing to meet the needs of its users as AI reduces the need for traditional BI. https://cdn.ttgtmedia.com/rms/onlineimages/ai_a205627811.jpg https://www.techtarget.com/data-technologies/news/366649895/As-AI-evolves-Tableau-talks-direction-ahead-of-Dreamforce Mon, 31 Aug 2026 15:14:00 GMT <p><i>AWS is ready for the ways agentic AI is changing database security needs.</i></p> <p><i>That's according to Eric Brandwine, a vice president and distinguished engineer for the tech giant and a longtime data security specialist.</i></p> <p><i>Security is a fundamental aspect of </i><a href="https://www.techtarget.com/data-technologies/news/366577632/Vector-search-and-storage-key-to-AWS-database-strategy"><i>AWS's database strategy</i></a><i>, Brandwine told TechTarget. Rather than being an afterthought, a feature that gets bolted on after a database is developed, security is built into AWS databases such as </i><a href="https://www.computerweekly.com/blog/CW-Developer-Network/Amazon-DynamoDB-now-supports-real-time-vector-search"><i>DynamoDB</i></a><i>, Aurora, RDS (Relational Database Service), ElastiCache, DocumentDB and Neptune.</i></p> <p><i>And it's always been that way.</i></p> <p><i>Agentic AI, meanwhile, is changing database security needs. Because agents can autonomously search and act on data at exponentially greater speed and scale than humans, they can find cracks in</i><i> database security that humans likely would have never discovered. They can even forge new vulnerabilities that, if not caught quickly, could expose an organization's proprietary information.</i></p> <p><i>In addition, because agents can operate so much faster than humans, they are enabling threat actors to </i><a href="https://www.computerweekly.com/news/366642503/AI-is-widening-the-asymmetry-between-attackers-and-defenders"><i>automate attacks</i></a><i>&nbsp;at higher volume than in the past.</i></p> <p><i>In response, </i><a href="https://www.techtarget.com/data-technologies/news/366644845/Agents-are-altering-data-security-needs-Oracle-responds"><i>Oracle</i></a><i> recently made some of its security capabilities available to customers at a significant discount. Similarly, database providers such as Microsoft and MongoDB added security capabilities specifically designed for agentic AI, while Snowflake improved its security capabilities after hackers used a former employee's password to </i><a href="https://www.techtarget.com/cybersecurity/news/366588655/Mandiant-Exposed-credentials-led-to-Snowflake-attacks"><i>steal customer</i><i> data</i></a><i>.</i></p> <p><i>AWS is likewise ready to both defend against attacks powered by agents as well as deploy agents to defend against nefarious activity, according to </i><i>Brandwine. And AWS has been seen since it first started offering database services, with agentic AI just the latest new technology to deploy and defend against. </i></p> <p><i>Eventually, there will be another, and with strict access controls and identity management, encryption while data is both at rest and in motion, and virtual private clouds to isolate networks, AWS has security capabilities built into its databases that are designed to stop any threat before it causes a significant breach.</i></p> <p><i>In a recent interview, Brandwine discussed AWS's database strategy, including the longstanding value it has placed on protecting</i><i> enterprise data and how its security capabilities are designed to shield against even threats powered by agents and other emerging technologies.</i></p> <p><i>In addition, he spoke about the benefits of </i><i>sufficient database security and the costs of insufficient database security, the new features that are being added to improve AWS's database security, and how agentic AI is driving </i><a href="https://www.computerweekly.com/microscope/feature/The-AI-cyber-security-challenge"><i>that evolution</i></a><i>.</i></p> <p><b>Editor's note</b>: <i>This Q&amp;A has been edited for clarity and conciseness</i>.</p> <p><b>Before the rise of agentic AI over the past couple of years, what was the typical security setup for a database?</b></p> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/brandwine_eric.jpg" alt="Eric Brandwine, vice president and distinguished engineer at AWS">Eric Brandwine </div> <p>Eric Brandwine: The database security landscape has been, and always will be, incredibly varied. At AWS, [Amazon president and CEO] Andy Jassy used to get on stage every year at re:Invent and say, 'Security is job zero.' We take that seriously. It is our goal to make AWS the most secure place for any customer to run any workload, and that extends to our database services. They have been designed, from day one, to provide a secure place for customers to place their most sensitive data.</p> <p>The purpose of a database is to host incredibly important data -- the lifeblood of an organization -- and <a href="https://www.techtarget.com/data-technologies/opinion/The-database-is-the-new-battleground-for-enterprise-AI">make it accessible to processes and now agents</a>, and if you don't do that securely, then you're going to have significant problems.</p> <p><b>Has agentic AI changed the security requirements for a database, and if so, how?</b></p> <p>Brandwine: I don't think that there are any new foundational problems due to agents. But organizations around the world have seen similar effects every time knowledge management -- information retrieval -- gets better. We've all always had search indices and internal document repositories, and every time the search engine improves, you discover that there are bunch of documents with the wrong permissions on them or that a set of people has access to information to information they shouldn't have had access to.</p> <p>Large language models (LLMs) and agents are the best information retrieval technology that humanity has ever had, and so we are finding across the industry that there is a renewed focus on making sure that the agent has access to everything that it needs -- an agent that doesn't have access to appropriate information is an agent that's ineffective – but <a href="https://www.techtarget.com/data-technologies/feature/How-agentic-AI-governance-tackles-data-security-challenges">what it needs and no further</a>. It's a problem that's been with us for as long as we've had databases.</p> <p><b>Agentic AI exponentially increases the volume and sophistication of the attacks they can make on databases – is that something AWS has had to address?</b></p> <p>Brandwine: We are seeing an increase in unwanted activity of certain kinds, and we believe it is attributable to agents. But fundamentally, agents are just a new technology. It is a thing that enables new things to happen, and they can be used by defenders as well as adversaries. At AWS, we have been making use of all the tools at our disposal, including agents, to better defend <a href="https://www.techtarget.com/data-technologies/feature/Bundesliga-delivering-insight-to-fans-via-AWS">our customers</a>.</p> <p>Picking a cloud provider is a long-term relationship. You should pick a provider that has a proven track record that you have faith in and will evolve with you as the future unfolds and will provide protection no matter what the next big technology is. When we talk about this with customers, databases are absolutely an important part of the conversation, but they are part of <a href="https://www.techtarget.com/cybersecurity/opinion/Security-highlights-from-AWS-reInvent-2025">the AWS picture as a whole</a>.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> The purpose of a database is to host incredibly important data -- the lifeblood of an organization -- and make it accessible to processes and now agents, and if you don't do that securely, then you're going to have significant problems. </figure> <figcaption> <strong>Eric Brandwine</strong>Vice president and distinguished engineer, AWS </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p><b>What is that picture as a whole, as it relates to security and defending against unwanted activity?</b></p> <p>Brandwine: We have AWS Private Cloud VPC with multiple layers of network protections and additional network telemetry, and we have our security services like Guard Duty where we do the monitoring for you to identify potentially unwanted activity. You have to look at the entire suite of services that a provider offers and evaluate that for your security needs.</p> <p><b>What can be the costs of insufficient database security, and conversely, what are the benefits of appropriate database security?</b></p> <p>Brandwine: The ideal security model is one where nothing you don't want ever happens, but everything that you do want happens seamlessly. That is the perfect model. We'll never exactly achieve that, but that is what we strive for. [In the ideal], you minimize your costs … and you maximize the amount of value that you can provide -- the innovation of your builders is supported and your <a href="https://www.techtarget.com/it-strategy/definition/business-process">business processes</a> are running without impact.</p> <p>The costs of poor database security are the exact opposite of this. In pretty much every application, data is the center of that application. It exists to manage, manipulate, query, and derive insight from. Almost always the core of that application is going to be some kind of database. One of the worst conversations you can have with your leadership is that you've had some kind of security issue and you can continue operating in an unknown state or you can take the application offline. Proactive investment and making sure you have the right mechanisms in place up front enable you to confidently ensure continued business operations.</p> <p><b>You mentioned that security has been integral to AWS' database strategy from the outset, but has agentic AI forced AWS to make any changes over the past couple of years?</b></p> <p>Brandwine: Agentic AI has impacted everyone's roadmap. We have features like <a href="https://www.techtarget.com/data-technologies/feature/Vector-search-now-a-critical-component-of-GenAI-development">vector search</a> in DynamoDB and [vector storage] in S3, but the security story for our databases is on the same trajectory as it's always been on.</p> <p>An agent fundamentally is just another actor in the system. It's something that can make API pulls, it's something that can issue queries, and from a security perspective, it doesn't matter what the source of those requests or queries is. We have to make sure that things the customer wants to enable are enabled, and that everything else is denied.</p> <p><b>We've been speaking very broadly about database security, but getting down to some details, what are some of AWS' key database security features?</b></p> <p>Brandwine: If you look at the suite of services that we offer, the different engines, customers want all of these different data management technologies and database engines. That can be open source engines like <a href="https://www.theserverside.com/tip/MySQL-vs-PostgreSQL-Compare-popular-open-source-databases">PostgreSQL</a> and <a target="_blank" href="https://mariadb.org/" rel="noopener">MariaDB</a>, or proprietary engines like Oracle and Microsoft SQL Server. Customers want us to support these engines, and they want new releases very quickly after they become available.</p> <p>That means we are constantly updating our database services. Databases themselves are not designed as security containers. The people that authored any of these databases -- Oracle, MySQL -- did not design them to hold an adversary inside the database, so we don't treat the database as a security container. We build multiple layers of security around that database.</p> <p><b>Can you give an example of what that looks like in a real-world situation?</b></p> <p>Brandwine: Last year at re:Invent, we shared that a couple of researchers from a company called Varonis managed to break out of their PostgreSQL instance. This is something that we strive to prevent, but we acknowledge is inevitable. … These researchers were able to break out of the database instance, but then our security measures kicked in, they were immediately detected, they were immediately stopped, and we were on the phone with them pretty quickly and asking what was going on.</p> <p>That's not really a feature of the database. It's just an inherent property of it. It's not an API that customers turn on. It's a thing that's a part of every database service that we offer, and … not something you bolt on afterward.</p> <p><b>What could have happened if AWS hadn't been so quick to detect the researchers – are there barriers in place that would have prevented them from going further beyond their PostgreSQL instance than they did?</b></p> <p>Brandwine: I can speak for hours about how we have done identity management and privilege management within our database services, ensuring that if someone does manage to gain even more access than those Varonis researchers managed to gain, that there aren't any credentials that they could get that would give them cross-tenant access. There <a href="https://www.techtarget.com/data-technologies/tip/How-AI-is-changing-data-protection">isn't any next hop</a> that they could take that could give them broad access to our customers' data.</p> <p><b>Since agents are designed to act autonomously, how is AWS empowering agents to act but limiting what they can access to ensure proper security?</b></p> <p>Brandwine: It is incredibly important that agents have credentials, but our rule is that no agent should ever get a credential that wasn't explicitly issued to it. The challenge that we've had with agents, and we're seeing this broadly across the industry, is that they have goal-seeking behavior. The agent determines through a process that to make a change, it has to delete the production database. Clearly, deleting the production database is <a target="_blank" href="https://www.youtube.com/watch?v=fVrw8V3iFLc" rel="noopener">a bad thing</a>. The agent will seek to get credentials that will delete the database. Making sure that the credentials that are issued to the agent are scoped to the activity that you've asked the agent to perform, and that the agent cannot get access to credentials, is incredibly important.</p> <p><b>As AWS' database strategy evolves, what is generally the impetus for change -- does it come from customer feedback, or is something else often the main motivator?</b></p> <p>Brandwine: We talk a lot about working backwards from customers. But Henry Ford has an interesting quote about this, which that if he'd listened to his customers, he would have built a faster horse. The point is that your customers … won't necessarily ask for the thing that we should build, so we have to innovate on their behalf. We have to talk to a whole bunch of customers and spot the pattern in what they're asking for, then come up with an idea that is not the next step forward, but is the thing that really answers the customers' needs.</p> <p><b>With many organizations now putting agents into production, trying to lower the cost of AI development and deployment is becoming a trend</b><b>. Is AWS doing things to help users keep their database spending under control?</b></p> <p>Brandwine: We have a broad sweep of cost management features. In fact, [we introduced] <a href="https://www.techtarget.com/data-technologies/news/366635663/Latest-AWS-data-management-features-target-cost-control">a number of them</a> at re:Invent last year.</p> <p>Another thing that's applicable here is that in <a href="https://www.theserverside.com/blog/Coffee-Talk-Java-News-Stories-and-Opinions/What-is-Amazon-Bedrock">Bedrock</a>, which is our model hosting service, we seek to provide customers with the broadest collection of models anywhere, and do so very promptly after a model is released. When a new model comes out, we do a few parameter changes to switch from one model to the new one so we can benchmark an application against the new model. What we find is that for any given application, there is not one right model. There are different workloads within an application. Some are complex and require advanced reasoning, and some just require parsing some human text and forming a response.</p> <p>Being able to <a href="https://www.techtarget.com/data-technologies/feature/SLM-vs-LLM-Rightsize-data-architecture-to-optimize-AI-use">select different models</a> and experiment rapidly to understand their strengths and weaknesses … has been a huge lever for us and our customers in managing the cost of their GenAI applications.</p> <p><b>What do you think will be the next big problem that database security measures will have to protect against?</b></p> <p>Brandwine: In the database security world, I don't know that there is a next big problem. In the agentic space, I think that <a href="https://www.techtarget.com/enterprise-software/feature/AI-feature-spend-is-the-new-software-cost-control-problem">the cost focus is increasing</a> broadly across the industry.</p> <p><b>As you plot AWS' database security strategy, what are some areas of focus?</b></p> <p>Brandwine: We pride ourselves on giving our customers choice. That means that we minimally constrain our customers. They can go innovate, they can have ideas that we never even contemplated and innovate on behalf of their customers and do all sorts of clever things. It's incredibly important that we enable that innovation. But, as a security professional, what I care about are the results that our customers achieve. Making sure that the vast majority of our customers have <a href="https://www.techtarget.com/cybersecurity/tip/Cloud-security-metrics-and-KPIs-A-CISOs-guide">a delightful day in the cloud</a> is our goal. We're going to continue the path that we've been on for years of relentlessly measuring ourselves, understanding the results that our customers are achieving, and figuring out how we can make those results better.</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> While the tech giant has always focused intently on protecting proprietary data, its database security capabilities are taking on new importance as agents increase vulnerabilities. https://cdn.ttgtmedia.com/rms/onlineimages/security_a210892891.jpg https://www.techtarget.com/data-technologies/feature/AWS-emphasizing-database-security-as-AI-powers-new-threats Mon, 24 Aug 2026 14:30:00 GMT <p>Data observability specialist Bigeye is lasering in on the cost of AI.</p> <p>Many enterprises are <a href="https://www.techtarget.com/enterprise-software/feature/AI-feature-spend-is-the-new-software-cost-control-problem">paying more attention to their spending</a> on AI development and management as they move AI initiatives out of the experimentation phase, when the primary focus is on ensuring that agents and other AI tools work in real-world situations, and into production environments.</p> <p>In response, Bigeye on Wednesday launched Cost Anomaly Detection within its Agent Trust Hub to automatically alert customers when agents and other data and AI consumers start spending more or less than their expected levels. Previously, a human would have had to manually monitor or audit an agent or user's behavior to catch sudden <a href="https://www.techtarget.com/it-strategy/feature/The-Meta-warning-When-AI-spending-becomes-a-liability">spending surges</a> or declines.</p> <p>Stewart Bond, an analyst at IDC, noted that his firm's research shows that two-thirds of the organizations it surveyed for a report released in July spent more on agentic AI than they budgeted.</p> <p>"Every dollar of wasted AI agent spend is a governance failure," Bond told TechTarget. "You cannot manage what you don't measure, and many organizations still aren't measuring agent spend in a way that catches problems as they happen, rather than a month later on an invoice."</p> <p>What is particularly valuable about Bigeye's new feature is that it surfaces not only AI spending increases but also decreases, he continued.</p> <p>"Catching an agent whose cost has quietly dropped to zero … is a different and often overlooked failure mode," Bond said. "Baselining against an entity's own history, with no budget model to configure upfront, also lowers the bar to actually turning this on, which matters given how many organizations already own cost dashboards they aren't fully using."</p> <p>Based in San Francisco, Bigeye is a <a href="https://www.techtarget.com/data-technologies/tip/The-5-pillars-of-data-observability">data observability</a> vendor that competes with fellow specialists such as Monte Carlo, along with broader-based data management providers including IBM and Informatica. In June, Bigeye launched Agent Trust Hub, a centralized location where users can observe agents, their activity and the <a href="https://www.techtarget.com/data-technologies/news/366646252/Context-is-king-as-agents-evolve-but-problems-persist">underlying data that informs them</a> to understand whether the AI tools can be trusted.</p> <section class="section main-article-chapter" data-menu-title="Addressing AI spending"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Addressing AI spending</h2> <p>The high cost of AI development and deployment is becoming a focal point for many enterprises. If spending spirals out of control, as it did for companies such as <a target="_blank" href="https://finance.yahoo.com/technology/ai/articles/canva-rare-startup-grew-fast-201427317.html" rel="noopener">Canva</a> and <a target="_blank" href="https://www.tradingview.com/news/reuters.com,2026:newsml_L4N4422AW:0-figma-s-upbeat-outlook-fails-to-stem-margin-worries-as-ai-costs-mount-shares-slump/" rel="noopener">Figma</a>, there can be significant consequences.</p> <p>Developing an agent, which involves model routing,&nbsp;context management and processing large amounts of data, can cost well over $100,000, according to IT consulting firm Triple Minds. Once in production, orchestration and continued calls to models and other tools add further expenses.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Every dollar of wasted AI agent spend is a governance failure. You cannot manage what you don't measure, and many organizations still aren't measuring agent spend in a way that catches problems as they happen, rather than a month later on an invoice. </figure> <figcaption> <strong>Stewart Bond</strong>Analyst, IDC </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>According to <a href="https://www.ey.com/en_us/newsroom/2026/07/ey-survey-c-suites-pivot-from-ai-adoption-to-unlocking-value-as-escalating-token-costs-trigger-fiscal-scrutiny">the fifth wave of EY's US AI Pulse Survey</a>, released in July, 82% of the surveyed senior leaders whose organizations are investing in AI report being concerned about their use of AI tokens and the cost of doing so.</p> <p>As a result, data and AI providers including AWS, Databricks, Microsoft and <a href="https://www.techtarget.com/data-technologies/news/366649521/Snowflake-targets-cost-of-AI-with-dynamic-model-routing">Snowflake</a> have all introduced features in recent months designed to help customers control their spending on AI.</p> <p>Bigeye is doing the same with Cost Anomaly Detection, with customer feedback motivating the vendor to develop the new feature, according to Bigeye CEO Eleanor Treharne-Jones.</p> <p>"As enterprises move AI agents from pilot into production, cost becomes hard to attribute -- a workflow's token usage can spike, or quietly drop to near-zero, and teams often don't notice until the bill lands or the agent stops delivering value," she told TechTarget. "We heard this directly and repeatedly from our own customers … and saw it validated more broadly across the market."</p> <p>Cost Anomaly Detection not only enables users to understand what they spent on AI, but when spending changes and why it does so. Each AI agent interaction that Bigeye observes has an estimated cost. When that cost changes beyond its expected range, Cost Anomaly Detection sends an alert that includes details about the AI actions that resulted in the disparity.</p> <p>In particular, Cost Anomaly Detection can discover when an agent slips into a retry <a href="https://www.techtarget.com/whatis/definition/loop">loop</a> or an unusually interactive conversation pattern with other applications, and costs suddenly increase. In addition, it can catch an agent that stops conversing with other tools because a credential expired or <a href="https://www.computerweekly.com/news/366643493/Avoid-expensive-AI-agents-with-these-five-design-imperatives">an integration failed</a>.</p> <p>"Cost Anomaly Detection is a real addition," Mike Leone, an analyst at Moor Insights &amp; Strategy, told TechTarget. "Most cost tools will catch an agent that suddenly triples its spend, and Bigeye does that. Catching an agent whose cost drops to zero is harder, and that usually means the agent stopped working. A falling bill looks like savings until somebody notices the work stopped and was never completed."'</p> <p>Monitoring for AI spending anomalies is not, however, a differentiator for Bigeye, he continued. But there is differentiation in the level of detail.</p> <p>"Most AI observability tools already track token spend," Leone said. "Where Bigeye does something less common is the attribution. It ties agent spend to the specific tables an agent queried, using the access records it already collects to enforce policy. It also works against older systems that predate the cloud warehouses."</p> <p>Bond likewise noted that, while the addition is valuable for Bigeye customers, <a href="https://www.techtarget.com/data-technologies/news/366630477/Monte-Carlos-Agent-Observability-targets-reliability-of-AI">Monte Carlo</a> also provides features that monitor data, code, AI models and infrastructure to analyze costs. Similarly, <a href="https://www.techtarget.com/data-technologies/news/366623394/Acceldata-launches-agentic-AI-powered-anomaly-detection">Acceldata</a> alerts users to cost overruns and performance degradation.</p> <p>However, Bigeye is distinguishing itself from competing vendors by integrating anomaly-based cost detection with existing data lineage and governance capabilities, Bond continued. With Cost Anomaly Detection, a change in the cost of an agent can be seen in conjunction with what data the agent accessed and whether that <a href="https://www.techtarget.com/data-technologies/opinion/Generative-AI-shines-spotlight-on-data-governance-and-trust">data can be trusted</a>.</p> <p>"The market's real gap isn't visibility anymore," he said. "It's turning visibility into enforceable outcomes, such as budget guardrails and automated rightsizing. That is likely where the real competitive separation will happen next, and it isn't fully solved by Bigeye or anyone else yet."</p> </section> <section class="section main-article-chapter" data-menu-title="Next steps"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Next steps</h2> <p>After launching the Agent Trust Hub and adding Cost Anomaly Detection over the past few months, Bigeye's focus for the remainder of 2026 is to add more capabilities that enable customers to understand and <a href="https://www.techtarget.com/searchenterpriseai/post/How-CIOs-should-architect-trust-in-AI-not-just-govern-it">trust their agentic AI tools</a>, according to Treharne-Jones.</p> <p>"We are focused on … deepening the trust and governance layer for agentic AI as more of these systems move into production, and expanding the platforms [to which] Agent Trust Hub connects to provide comprehensive visibility across the AI ecosystem," she said.</p> <p>Bond advised Bigeye to advance tools such as Cost Anomaly Detection from merely alerting users to anomalies to automatically resolving issues such agentic AI cost overruns.</p> <p>"Vendors need to shift from reporting features toward real budget guardrails and automated rightsizing, since the tooling for visibility already exists broadly in the market and isn't closing the value gap on its own," he said.</p> <p>In addition, adding capabilities that <a href="https://www.techtarget.com/ai/tip/The-best-AI-governance-tools-and-platforms-in-2026">trace every decision</a> and action made by AI tools would be valuable for Bigeye users, Bond continued.</p> <p>"IDC found that maintaining a complete auditable record of agent decisions is the fastest-rising pain point in agent operations, even as raw security concerns decline with maturity," he said. "A natural next step for Cost Anomaly Detection would be to tie anomalies not just to conversations but to a defensible audit trail of what the agent decided and why."</p> <p>Leone likewise suggested that Bigeye integrate policy enforcement with anomaly detection. He noted that the vendor already has a policy layer that sits between AI agents and data. Adding a feature that <a href="https://www.techtarget.com/ai/tip/Why-businesses-need-an-AI-agent-kill-switch">stops agents</a> from spending when they hit a certain level is a natural next step.</p> <p>"Flagging that an agent spent five times its normal amount yesterday helps, [but] cutting the agent off when it crosses twice its normal amount helps more," Leone said.</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> With agentic AI costing more than many organizations expected, the data observability specialist's latest alerts users when the expense of an agent suddenly changes. https://cdn.ttgtmedia.com/rms/onlineimages/money_g972609480.jpg https://www.techtarget.com/data-technologies/news/366649478/Data-observability-specialist-Bigeye-puts-focus-on-AI-spend Wed, 19 Aug 2026 14:35:00 GMT <p>Snowflake is doing something about the high cost of developing and managing agents and other AI applications.</p> <p>Building an agent requires workflows such as model routing, <a href="https://www.techtarget.com/data-technologies/news/366646252/Context-is-king-as-agents-evolve-but-problems-persist">context management</a>, governing tool calls, batch data processing and network orchestration that can sometimes add up to well over $100,000, according to IT consulting firm Triple Minds. As a result, it is often significantly more expensive to develop AI tools than to create the traditional data products that have historically been the primary sources enterprises used to inform decisions and manage business processes.</p> <p>With many organizations now <a target="_blank" href="https://www.gartner.com/en/newsroom/press-releases/2026-1-15-gartner-says-worldwide-ai-spending-will-total-2-point-5-trillion-dollars-in-2026" rel="noopener">spending far more</a> to build AI tools than they did to develop data products, Snowflake on Tuesday unveiled dynamic model routing in its Cortex AI Gateway control layer for AI.</p> <p>Dynamic model routing aims to reduce wasted spending on using bigger, more expensive models than are needed to inform many agents and instead automate model selection to strike the optimal balance between the quality of the model and <a href="https://www.techtarget.com/enterprise-software/feature/AI-feature-spend-is-the-new-software-cost-control-problem">the cost of the project</a>.</p> <p>"These capabilities are highly significant because they attack the operational overhead and suboptimal selection of models," William McKnight, president of McKnight Consulting, told TechTarget. "By automatically matching task complexity to the optimal model, dynamic routing eliminates the need for developers to rebuild their application infrastructure every time a new model is released."</p> <p>Beyond dynamic model routing, Snowflake revealed that it plans to add models from <a href="https://www.techtarget.com/whatis/feature/DeepSeek-explained-Everything-you-need-to-know">DeepSeek</a> and Z.ai to <a href="https://www.techtarget.com/data-technologies/news/366625218/Snowflake-continues-to-add-AI-boost-Cortex-capabilities">its Cortex AI development environment</a>. The vendor already enables access to models from Anthropic, Google, Mistral, OpenAI and SpaceX. More model choices better enable users to strike the optimal balance between cost and performance.</p> <p>Dynamic model routing and access to Z.ai's GLM-5.3 are not yet in private preview, while access to DeepSeek-V4-Flash is in private preview.</p> <section class="section main-article-chapter" data-menu-title="Customers want cost control"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Customers want cost control</h2> <p>AWS in December made cost control the focal point of <a href="https://www.techtarget.com/data-technologies/news/366635663/Latest-AWS-data-management-features-target-cost-control">its data management initiatives</a> given that AI development demands massive data workloads. Databricks and Microsoft also offer capabilities aimed at helping customers control their spending on AI.</p> <p>Now, Snowflake is similarly addressing the cost of AI development and deployment.</p> <p>As more enterprises move AI pilots into production, they're paying more attention to spending than they did when they were just trying to prove that something worked, according to Sanjeev Mohan, founder and principal analyst of analyst firm SanjMo. That is the reason data and AI providers are beginning to prioritize cost control.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Companies run AI across many apps and agents, and the default habit of sending every request to the most powerful frontier model turns out to be enormously wasteful, since most tasks don't need that horsepower. </figure> <figcaption> <strong>Sanjeev Mohan</strong>Founder and principal analyst, SanjMo </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>"Companies run AI across many apps and agents, and the default habit of sending every request to the most powerful frontier model turns out to be enormously wasteful, since most tasks don't need that horsepower," he told TechTarget.</p> <p>For example, Canva was <a target="_blank" href="https://finance.yahoo.com/technology/ai/articles/canva-rare-startup-grew-fast-201427317.html" rel="noopener">forced to cut its growth forecast</a> from 30% to 20% because it relied too heavily on frontier models that proved more costly than expected, Mohan continued.</p> <p>"The analyst read is that AI inference breaks the near-zero marginal cost that made software so profitable. … Managing spend across models is now worth billions," he said.</p> <p>McKnight likewise suggested that as enterprises move more AI pilots into production, cost is becoming <a href="https://www.techtarget.com/it-strategy/feature/What-Big-Techs-AI-spending-means-for-your-IT-budget">a greater concern</a>, and vendors must respond to keep their customers.</p> <p>"Enterprises are under pressure to transition speculative AI pilots into ROI-generating production and contain runaway cloud compute costs and the extreme engineering overhead of multi-vendor stacks," he said.&nbsp;"To prevent customers from abandoning these initiatives, database and orchestration vendors like Snowflake are prioritizing predictable pricing models that make AI development financially sustainable."</p> <p>Customers, in fact, provided Snowflake with the impetus to develop dynamic model routing and expand the selection of models its users can choose from when building AI tools, according to Pavan Pothukuchi, the vendor's director of product for AI and machine learning.</p> <p>He noted that with customers "becoming more rigorous about the economics" as AI <a target="_blank" href="https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html?id=us:2ps:3gl:aisgm26:awa:CONS:em:K0218784:012626:kwd-430833501819:195648817121:794247818306::&amp;gclsrc=aw.ds&amp;gad_source=1&amp;gad_campaignid=23269751971&amp;gclid=CjwKCAjwhZDUBhBGEiwAbi5bjpnKWPhwQ_x2iRpG4nnYjNKAl3JOosCPrbhIeSMsOTahBvicfzbySRoCS8MQAvD_BwE" rel="noopener">moves from experimentation to production</a>, Snowflake saw an opportunity to help them reduce spending by optimizing model selection.</p> <p>"Customers weren't necessarily asking us specifically for dynamic model routing, but they were asking us to help improve the economics and performance of their AI deployments," Pothukuchi said. "Dynamic model routing is a natural response to that … so customers can focus on the outcomes they want from AI."</p> </section> <section class="section main-article-chapter" data-menu-title="Addressing AI spending"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Addressing AI spending</h2> <p>Snowflake introduced Cortex AI Gateway in July 2025 to provide administrators with a single location to govern connections and interactions between agents, intelligently route workloads, and optimize AI use. Within Cortex AI Gateway, users can view <a href="https://www.techtarget.com/whatis/definition/token">token</a> usage and costs, and establish spending limits for agents and other AI tools so they don't unexpectedly drive up costs.</p> <p>The addition of dynamic model routing helps customers control their AI spending by automating the selection of the models that agents are built upon.</p> <p>Snowflake noted that using the same model for every task can drive up costs since certain models are better suited for an agent's specific task than others. For example, one model may be better suited for informing <a href="https://www.techtarget.com/enterprise-software/feature/Early-adopters-of-agentic-AI-for-customer-service-offer-advice">customer service agents,</a> while another may be better suited for <a href="https://www.computerweekly.com/feature/Forrester-Managing-supply-chain-volatility-with-agentic-AI">supply chain optimization agents</a>.</p> <p>Dynamic model routing automatically funnels work that necessitates deep reasoning to frontier models while directing lower complexity tasks to less expensive models. In addition, routing decisions update as models and their pricing change, so users don't have to rebuild agents and other AI applications to lower spending.</p> <p>"Dynamic model routing directly addresses the primary bottleneck of enterprise AI, which is the operational burden of manual model selection at scale and automatically matching each workload to the optimal AI model based on quality, speed, and cost," McKnight said.</p> <p>Mohan likewise noted that the new feature directly addresses a substantial problem many enterprises face.</p> <p>"The routing … removes a real operational chore, and does it automatically inside the platform where the data already lives," he said.</p> <p>While dynamic model routing addresses customers that want to control costs, it does not distinguish Snowflake from its main competitors, according to Mohan. He noted that Microsoft offers Model Router in <a href="https://www.techtarget.com/ai/news/366616024/Microsoft-intros-Azure-AI-Foundry-for-building-AI-apps">AI Foundry</a>, Databricks provides routing through Unity AI Gateway, and vendors including OpenRouter and Vercel automate model routing.</p> <p>However, its deep integration with Snowflake's <a href="https://www.techtarget.com/data-technologies/news/366643795/Snowflake-barrage-adds-more-AI-development-analysis-tools">broader data and AI platform</a>, including tools such as CoCo and CoWork, along with governance capabilities such as role-based access, does provide differentiation.</p> <p>"What's different about Snowflake's version is where it sits, routing inside Snowflake's governed boundary," Mohan said. "For a Snowflake-centric enterprise, routing that never moves data outside governance and attributes spending to the right team is the distinction, not the routing algorithm itself."</p> <p>Meanwhile, Snowflake's overall cost of building and managing AI workloads is competitive, according to McKnight. Capabilities such as real-time concurrent LLM serving with a 0% error rate, optimized compute sizing, and Snowflake <a href="https://www.techtarget.com/it-infrastructure/definition/What-is-container-management-and-why-is-it-important">Container Services</a> aid its price-performance ratio.</p> <p>"Against the native services of major cloud providers, Snowflake is a highly competitive, well-balanced option for search," McKnight said.</p> <figure class="main-article-image full-col" data-img-fullsize="https://www.techtarget.com/rms/onlineimages/gen_ai_costs_beyond_development-h.png"> <img data-src="https://www.techtarget.com/rms/onlineimages/gen_ai_costs_beyond_development-h_mobile.png" class="lazy" data-srcset="https://www.techtarget.com/rms/onlineimages/gen_ai_costs_beyond_development-h_mobile.png 960w,https://www.techtarget.com/rms/onlineimages/gen_ai_costs_beyond_development-h.png 1280w" alt="AI costs extend beyond the initial feature price. Change management, infrastructure, model run costs, governance and ongoing optimization all affect the real cost of enterprise AI. " data-credit="TechTarget"> <figcaption> <i class="icon pictures" data-icon="z"></i>With model routing among the reasons AI development and deployment costs are spiraling for some organizations, Snowflake introduced a feature that automates model selection to optimally balance cost and performance. </figcaption> <div class="main-article-image-enlarge"> <i class="icon" data-icon="w"></i> </div> </figure> </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Looking ahead</h2> <p>Over the final months of 2026, Snowflake's plans include improving how the pieces that comprise <a href="https://www.techtarget.com/ai/tip/Agentic-AI-workflows-Trends-examples-and-best-practices">the AI stack</a> work together, according to Pothukuchi. Regarding dynamic model routing, he added that making the feature more adaptive will be a priority.</p> <p>"We want to get increasingly sophisticated at understanding what a particular customer is trying to accomplish and selecting the intelligence best suited to that task, while continuously optimizing for quality and efficiency," Pothukuchi said.</p> <p>Mohan suggested that Snowflake could build on dynamic model routing by adding cost control tools that go beyond routing to include generating and acting on their own recommendations.</p> <p>"The natural next step is … moving from control to active guidance, which is exactly the muscle companies like Canva had to build by hand," he said.</p> <p>McKnight, meanwhile, advised Snowflake to take steps to address the complexity of <a href="https://www.techtarget.com/data-technologies/news/366641473/Snowflake-broadens-open-source-embrace-ups-Iceberg-support">its platform</a> and opaque billing that can catch customers off guard.</p> <p>Specifically, it could turn its open-source pg_lake extension into a fully managed rival of <a href="https://www.techtarget.com/data-technologies/news/366638723/Databricks-launches-PostgreSQL-Lakebase-to-aid-AI-developers">Databricks' Lakebase,</a> natively integrate an open semantic layer with Cortex AI and its Horizon Catalog to securely govern data mesh architectures, and add AI-driven <a href="https://www.techtarget.com/ai/tip/How-to-apply-FinOps-to-optimize-agentic-AI-costs">FinOps controls</a> along with micro-concurrency scaling to make spending more predictable.</p> <p>"To continue holding its prime position in the market, Snowflake must focus on eliminating architectural complexity and billing anxiety," McKnight said.</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> With spending on AI spiraling out of control for many organizations, the vendor's latest new feature automates model selection to balance cost and performance. https://cdn.ttgtmedia.com/rms/onlineimages/money_g1050046190.jpg https://www.techtarget.com/data-technologies/news/366649521/Snowflake-targets-cost-of-AI-with-dynamic-model-routing Tue, 18 Aug 2026 09:07:00 GMT <p>Databricks keeps raising funding.</p> <p>After revealing in July that it was in the process of securing new financing, Databricks on Thursday closed on $5 billion in venture capital funding at a rate that values the company at $190 billion and brings its total financing over $32 billion.</p> <p>But after raising $5 billion or more in four separate funding rounds -- <a href="https://www.techtarget.com/data-technologies/news/366617434/Databricks-record-10-billion-funding-round-to-fuel-growth">including $10 billion in December 2024</a> -- the question isn't what the data management and AI vendor should do with the money or whether it can raise more. Instead, it is what Databricks should do next.</p> <p>While many data management and analytics vendors are struggling to attract investors, there seems to be an endless supply of private capital willing to invest in Databricks. Beyond rounds exceeding $5 billion, the vendor raised $2 billion in debt financing in February and added $1 billion or more three additional times.</p> <p>However, an eventual initial public stock offering would be more advantageous for Databricks than continuing to raise private funding, according to Michael Ni, an analyst at Constellation Research.</p> <p>"Staying private has allowed Databricks flexibility to ignore public scrutiny on its profitability and predictability, but a successful public IPO could provide a growth flywheel," he told TechTarget.</p> <p>Specifically, if the public markets respond positively to Databricks' <a href="https://www.techtarget.com/data-technologies/news/366644813/Databricks-intros-new-Genie-data-management-tools-to-aid-AI">continued additions</a> of &nbsp;data and AI stack components, a healthy stock price would give the vendor greater flexibility to expand than raising private funding, he continued. In addition, going public -- and the need to respond to public-market sentiments -- would make it easier for the CIOs of potential customers to make Databricks part of their AI and <a href="https://www.techtarget.com/data-technologies/definition/What-is-data-architecture-A-data-management-blueprint">data architectures</a>.</p> <p>"The value in an IPO would go beyond the cash raised, enabling Databricks to move from a successful private platform to a consolidator and long-term standard for the enterprise AI stack," Ni said.</p> <p>However, when an IPO happens depends, to some degree, on external factors such as the <a target="_blank" href="https://www.ey.com/en_us/insights/ipo/ipo-market-trends" rel="noopener">market conditions for major IPOs</a>. Also in question is how significantly an IPO would benefit Databricks when the company is seemingly able to raise funding whenever it desires to fuel technological innovation, acquisitions and other initiatives.</p> <section class="section main-article-chapter" data-menu-title="Databricks is different"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Databricks is different</h2> <p>Databricks is far from the only data management vendor that has expanded over the past few years to provide a data foundation for AI development.</p> <p>Rival <a href="https://www.techtarget.com/data-technologies/news/366643795/Snowflake-barrage-adds-more-AI-development-analysis-tools">Snowflake</a> and hyperscale cloud providers <a href="https://www.techtarget.com/data-technologies/news/366644853/AWS-latest-to-introduce-context-layer-for-agentic-AI">AWS</a>, <a href="https://www.techtarget.com/data-technologies/news/366641929/Google-unveils-data-cloud-purpose-built-for-agentic-AI">Google Cloud</a> and <a href="https://www.techtarget.com/data-technologies/news/366643955/Microsoft-boosts-Fabric-to-make-it-a-foundation-for-AI">Microsoft</a> have all similarly added capabilities that enable customers to connect AI tools with proprietary data. In addition, database vendors such as MongoDB and Couchbase, niche specialists including Alation and Informatica, and analytics providers such as Qlik and Tableau have all added tools to enable AI development <a href="https://www.techtarget.com/data-technologies/news/366646252/Context-is-king-as-agents-evolve-but-problems-persist">informed by relevant, high-quality data</a>.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> The value in an IPO would go beyond the cash raised, enabling Databricks to move from a successful private platform to a consolidator and long-term standard for the enterprise AI stack. </figure> <figcaption> <strong>Michael Ni</strong>Analyst, Constellation Research </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>Databricks' valuation of $190 billion, however, dwarfs Snowflake's market capitalization of $114 billion, despite Snowflake's stock price more than doubling since April. And while Databricks continues to raise billions of dollars with each funding round, vendors such as Confluent, Domo, Dremio, DBT Labs and Fivetran have either sought buyers or merged with peers with funding difficult for data and analytics vendors to raise since <a href="https://www.techtarget.com/data-technologies/news/252520740/Tech-stock-sell-off-signals-tough-times-for-data-vendors">tech stocks plummeted in 2022</a>.</p> <p>Venture capitalists simply view Databricks differently than they view most other data and AI providers, according to Sanjeev Mohan, founder and principal of analyst firm SanjMo.</p> <p>"Investors see Databricks becoming the single consolidated platform for a wide span of data and AI use cases," he told TechTarget. "Databricks is different in that it actually keeps absorbing adjacent categories onto one architecture rather than bolting on a feature and calling it a platform."</p> <p>In addition, venture capitalists have faith in Databricks CEO Ali Ghodsi, Mohan continued.</p> <p>"Investors find their CEO to be a force of nature who has tremendous ambition and grit to build out the platform," he said.</p> <p>One of those investors is Benjamin Black, <!--StartFragment--><span data-olk-copy-source="MessageBody">co-founder and CIO of Powerlaw Capitals Group, which holds exposure to Databricks through Powerlaw Corp. </span></p> <p>Numerous factors have led investors to flock to Databricks while only minimally funding some of its smaller competitors, according to Black. In particular, he noted that its execution at scale -- surpassing $7 billion in annual recurring revenue and year-over-year revenue growth of approximately 80% -- are significant factors in the venture capital community's <a href="https://www.techtarget.com/data-technologies/news/366617350/Record-funding-round-reflects-Databricks-differentiation">continued interest in Databricks</a>.</p> <p>"Databricks is popular because it has demonstrated unusually strong execution at enormous scale just as enterprises are trying to turn AI from a promising experiment into a core business capability," Black said.</p> <p>Strong growth at scale and technology execution -- a platform designed to enable users to successfully develop AI tools -- makes Databricks not merely a company investors want as part of their portfolio, but one they want to get in on early, he continued.</p> <p>"It has the revenue, growth, and institutional maturity to list when it chooses," Black said. "That combination is precisely why investors want exposure now rather than waiting for a ticker symbol.</p> <p>Beyond how investors view Databricks, which is as much a bet on growth potential and financial performance than a recognition of technological prowess, Databricks has distinguished itself from a technological perspective as well, according to David Menninger, an analyst at ISG Software Research.</p> <p>Significant features Databricks has developed over the past few years include Unity Catalog for governing data and AI, Mosaic AI for developing generative and agentic AI, <a href="https://www.techtarget.com/data-technologies/news/366638723/Databricks-launches-PostgreSQL-Lakebase-to-aid-AI-developers">Lakebase</a> to provide a PostgreSQL database foundation for AI initiatives, <a href="https://www.techtarget.com/data-technologies/news/366625695/Latest-Databricks-tools-use-AI-to-simplify-AI-development">Agent Bricks</a> to automate aspects of AI development, and its Genie natural language interface.</p> <p>"Databricks has executed well," Menninger said. "Investors are primarily interested in high-growth companies. Databricks has been growing rapidly for many years."</p> <p>ISG groups Databricks with 12 other data and AI platform providers, he continued. Peers include fellow data platform providers Cloudera, Snowflake and Teradata, <a href="https://www.techtarget.com/it-infrastructure/definition/What-is-hyperscale-cloud-Computing-and-data-center-uses-explained">hyperscale cloud</a> vendors AWS, Google Cloud and Microsoft, and other broad-based technology providers such as IBM, Oracle and SAP.</p> <p>"We rate Databricks as an overall leader among this group, which means we place them in the top three," Menninger said.</p> </section> <section class="section main-article-chapter" data-menu-title="Private vs. public"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Private vs. public</h2> <p>In theory, Databricks could stay private for the foreseeable future.</p> <p>"Databricks has had virtually no financial reason to face the public markets because the private markets are treating them so favorably," William McKnight, president of McKnight Consulting, told TechTarget.</p> <p>Similarly, Mohan noted that private investors have been so willing to keep Databricks flush with cash that the vendor has the luxury of waiting until the ideal time to <a href="https://www.techtarget.com/data-technologies/news/366636532/Databricks-adds-4B-funding-round-IPO-could-be-next">explore an IPO</a>.</p> <p>"The private market has given Databricks everything an IPO would, without the constraints," he told TechTarget.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Private secondary sales only stretch so far. A public market lets employees and early backers actually cash out, unlike private market. That is good for employee retention and morale. </figure> <figcaption> <strong>Sanjeev Mohan</strong>Founder and principal analyst, SanjMo </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>Eventually, however, the benefits of going public outweigh those of remaining private, Mohan continued.</p> <p>For example, each private funding round dilutes the value that existing investors hold in Databricks. In addition, public companies have the transparency and credibility that can attract large enterprise and government customers.</p> <p>"Private secondary sales only stretch so far," Mohan said. "A public market lets employees and early backers actually cash out, unlike the private market. That is good for employee retention and morale."</p> <p>McKnight likewise noted that the <a href="https://www.investopedia.com/ask/answers/021015/what-difference-between-ipo-and-private-placement.asp">benefits of going public</a> ultimately outweigh those of continuing to raise private capital. Specifically, he noted that an IPO would enable Databricks to have access to liquid, market-validated funds for future acquisitions, stock to attract and retain elite talent, access to cheap debt and secondary capital markets, and SEC-audited transparency.</p> <p>"Even with unlimited private capital, going public provides advantages that private funding cannot replicate," McKnight told TechTarget.</p> <p>That said, raising more private funding remains a possibility for Databricks.</p> <p>Black noted that a company such as Databricks can theoretically remain private indefinitely, as long as it can continue to raise capital, its employees and early investors can get liquidity without the company going public, and it can demonstrate its financial health without public filings.</p> <p>"Private capital has gotten deep enough, with the largest funds able to write enormous checks, that a company can keep raising privately for a long time without ever needing the public markets," Black said.</p> <p>In fact, it's possible that Databricks and other companies,such as <a href="https://www.tradingview.com/symbols/NASDAQ-STRIPE/">Stripe,</a> that are able to raise massive amounts of funding, could create a new class of "forever private" companies, he continued.</p> <p>"Databricks and Stripe get all primary and secondary liquidity they need in the private markets," he said, noting that they can innovate without having to meet the expectations of public markets and each has already made the type of acquisitions that public stock is often used to fund. "If the private markets stay so accommodating for the best mega-privates, why go public at all?"</p> <p>Menninger likewise noted that Databricks -- in theory -- could keep raising funding and remain private. He pointed out that IPOs typically fund expansion, and Databricks has been able to expand by <a href="https://www.techtarget.com/data-technologies/news/366623864/Databricks-adds-Postgres-database-with-1B-Neon-acquisition">making numerous acquisitions</a> and investing in product development.</p> <p>"If you have access to the capital you need, it’s much better not to have to face the quarterly scrutiny of the public markets. It can really distract from the day-to-day operations of the business," Menninger said.</p> <p>However, the greater likelihood is that Databricks will eventually go public, he continued, noting that benefits such as the public disclosure of financials attracting new customers, equity given to employees having more value when a company is public, and more easily funded acquisitions could outweigh the benefits of remaining private.</p> <p>"Investors need to be able liquidate their investments at some point," Menninger said. "As long as there are other investors willing to step in, they can potentially go on like this forever, but that's highly unlikely. Even if they reach the point where they don't need additional capital, the existing investors will want to be able to close out their positions."</p> </section> <section class="section main-article-chapter" data-menu-title="If not now, when?"> <h2 class="section-title"><i class="icon" data-icon="1"></i>If not now, when?</h2> <p>Databricks will eventually go public, according to Ghodsi.</p> <p>"We will go public, I promise," he said during an interview with CNBC on Thursday following the closing of Databricks' $5 billion funding round.</p> <p>But the timing for an IPO has to be right, Ghodsi emphasized.</p> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/ghodsi_ali.jpg" alt="Databricks CEO Ali Ghodsi">Ali Ghodsi </div> <p>The stock market, though there have been significant IPOs in 2026, has been volatile, <a target="_blank" href="https://www.google.com/search?q=dow+jones&amp;oq=dow+jones&amp;gs_lcrp=EgRlZGdlKgYIABBFGDkyBggAEEUYOTIICAEQ6QcY_FXSAQgxODE2ajBqNKgCAbACAQ&amp;sourceid=chrome&amp;ie=UTF-8" rel="noopener">trending generally up</a> but lurching more in response to economic indicators, major companies' earnings filings and other reports.</p> <p>"The markets have been really wobbly," Ghodsi told CNBC. "For three months, everything is down and it's going to be terrible. And then the other day, it's amazing. There was one great [announcement] by one software company and the whole market is up 10%, and then the next day everything is down 10%. It's just very turbulent times."</p> <p>Beyond market unpredictability, the final months of 2026 may not be a good time for Databricks to go public because other major tech companies are also planning to go public and a Databricks IPO could get overshadowed.</p> <p>Already in 2026, <a href="https://www.techtarget.com/ai/news/366644478/SpaceX-IPO-aims-for-AI-and-orbital-data-centers">SpaceX</a> pulled off the largest IPO in history when it raised $75 billion. Just as Databricks was announcing its latest funding round, reports surfaced that <a href="https://www.techtarget.com/ai/news/366624572/Anthropic-intros-next-generation-of-Claude-AI-models">Anthropic</a> may be planning an IPO in October that could surpass SpaceX's. In addition, <a href="https://www.techtarget.com/ai/news/366628814/OpenAIs-GPT-5-is-out-Where-it-shines-and-where-it-doesnt">OpenAI</a> has already filed initial paperwork for an IPO.</p> <p>Those IPOs could take both potential investors and the infrastructure required to go public -- the investment firms that manage such undertakings -- away from Databricks.&nbsp;</p> <p>"Ghodsi has been open that he views this as a poor year to go public with SpaceX, OpenAI, and Anthropic all crowding the IPO conversation, and he's pointed to a possible window in 2027," Mohan said.</p> <p>One thing Databricks does not have to worry about is missing its opportunity to go public, which happened to numerous data management and analytics providers in 2022.</p> <p>Vendors including Qlik, <a href="https://www.techtarget.com/data-technologies/news/252477178/ThoughtSpot-IPO-could-be-coming-after-vendor-adds-first-CFO">ThoughtSpot</a> and <a href="https://www.techtarget.com/data-technologies/news/252516835/Pyramid-Analytics-raises-120M-IPO-could-be-next">Pyramid Analytics</a> spoke openly about IPOs, with Qlik even <a href="https://www.techtarget.com/data-technologies/news/252511695/Qlik-planning-an-IPO-files-application-with-the-SEC">filing its initial paperwork</a>. But after tech stocks plummeted in the spring of 2022, the technology landscape changed toward the end of the year, with AI becoming the dominant focus. Since then, niche data management and analytics vendors have had to reposition themselves and have largely lost the favor of the investment community.</p> <p>Qlik, ThoughtSpot and Pyramid remain private.&nbsp;</p> <p>"Databricks is delaying because public markets are already giving it public company resources without public company constraints," Ni said. "Databricks is using the financial flexibility to drive growth rather than prove quarterly predictability. An IPO will become more necessary when the need for employee liquidity, acquisition currency and public market permanence outweighs the benefits of flexibility."</p> </section> <section class="section main-article-chapter" data-menu-title="The outlook"> <h2 class="section-title"><i class="icon" data-icon="1"></i>The outlook</h2> <p>While an IPO is all but inevitable, and before going public Databricks remains able to attract large amounts of funding to fuel seemingly any growth initiative, technology needs to remain the vendor's main priority, according to McKnight.</p> <p>"My focus, and my clients' focus, is less on the financial headlines&nbsp;and much more on the core architectural and product execution&nbsp;they deliver to the enterprise," he said.</p> <p>For example, the <a href="https://www.techtarget.com/enterprise-software/feature/AI-feature-spend-is-the-new-software-cost-control-problem">high cost of AI development</a> is an ongoing concern for many enterprises, in some cases preventing companies from building agents and other applications. In addition, navigating Databricks to develop data and AI products can be complicated. Meanwhile, after acquiring a spate of companies -- including its purchase of PostgreSQL database innovator <a href="https://www.techtarget.com/data-technologies/news/366649200/Databricks-Electric-acquisition-adds-embeddable-PostgreSQL">ElectricSQL</a> this week -- there are still tools that need to be integrated with Databricks' existing platform before customers can take full advantage.</p> <p>"They&nbsp;could&nbsp;neutralize cost opacity, deliver a true zero-administration experience,&nbsp;provide prebuilt AI templates within Agent Bricks&nbsp;and see about&nbsp;integrating [acquisitions such as] PostgreSQL Lakebase (Neon)&nbsp;and Tabular under Unity Catalog," McKnight said, noting that some competitors deliver faster execution and lower costs across customers data estates.</p> <p>"Additionally, Databricks suffers from severe competitive multi-year ingestion pipeline degradation&nbsp;and critical autoscaling endpoint failures under real-time concurrent inference," he added.</p> <p>Mohan similarly suggested that while a well-timed IPO rather than another funding round is essential, so is ensuring that the many new features Databricks has <a href="https://www.techtarget.com/data-technologies/news/366637142/New-Databricks-tool-aims-to-up-agentic-AI-response-accuracy">developed</a> and <a href="https://www.techtarget.com/data-technologies/news/366588032/Databricks-1B-plus-Tabular-acquisition-adds-Iceberg-support">acquired</a> work as intended, and do so in an integrated manner that enables customers to build the data and AI tools they desire.</p> <p>"Proof that their expansion is actually working," he said when asked what he'd like to see from Databricks over then next year. "There is no guarantee that their forays into marketing and security -- or even Lakebase -- will appeal to buyers who may be reluctant to put all their eggs in one basket. I'd like to see them prove these expansions merge into one coherent product rather than a widening surface."</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> Experts note that the benefits of going public ultimately outweigh those of staying private, even with the vendor seemingly able to raise unlimited financing. https://cdn.ttgtmedia.com/rms/onlineimages/money_g1222040206.jpg https://www.techtarget.com/data-technologies/news/366649434/IPO-inevitable-for-Databricks-after-adding-5B-in-funding Fri, 14 Aug 2026 14:43:00 GMT <p>Databricks is going Electric.</p> <p>Though not as historic as Bob Dylan's move away from the acoustic guitar at the 1965 Newport Folk Festival, Databricks on Aug. 11 revealed the acquisition of startup ElectricSQL to add new <a href="https://www.theserverside.com/tip/MySQL-vs-PostgreSQL-Compare-popular-open-source-databases">PostgreSQL database capabilities</a>.</p> <p>Financial terms of the deal were not disclosed. Electric raised $600,000 in pre-seed funding in 2022.</p> <p>Databricks first added PostgreSQL database capabilities with its May 2025 <a href="https://www.techtarget.com/data-technologies/news/366623864/Databricks-adds-Postgres-database-with-1B-Neon-acquisition">acquisition of Neon</a>, which Databricks has subsequently transformed into the core of <a href="https://www.techtarget.com/data-technologies/news/366638723/Databricks-launches-PostgreSQL-Lakebase-to-aid-AI-developers">its Lakebase platform</a>. The acquisition of Electric allows Databricks to extend PostgreSQL database capabilities beyond data lakehouses to edge devices, including agents, additional AI applications and mobile devices.</p> <p>Perhaps the biggest benefit of the acquisition is that Databricks will enable agents to more quickly and easily fix themselves when they take missteps, given that the data they need to do so will be more readily available than before, according to Mike Leone, an analyst at Moor Insights &amp; Strategy.</p> <p>"[The acquisition] definitely matters, but it's more of a specific fix than a big swing," he told TechTarget, adding that agents need to keep notes as they operate to see when they go wrong and correct any errors. "Electric shrank Postgres down small enough to run beside the agent instead of across a network. That makes it fast, and it lets the agent undo a step cleanly instead of leaving a half-finished mess."</p> <p>Stephen Catanzano, an analyst at Omdia, a division of TechTarget, called Databricks' acquisition of Electric significant. Like Leone, he noted that it addresses a specific need in agentic AI development and deployment.</p> <p>"The acquisition addresses a critical infrastructure gap in the emerging world of agentic applications, where AI agents need both ultra-low latency access to local data in sandboxed environments and real-time synchronization with centralized systems," he told TechTarget. "Electric … extends Databricks' existing Postgres capabilities from the centralized lakehouse to the edge where agents actually execute."</p> <p>San Francisco-based Databricks' purchase of Electric is the latest in a series of acquisitions Databricks has made over the past few years to accelerate its expansion beyond data management <a href="https://www.techtarget.com/data-technologies/news/366633904/New-Databricks-tools-target-successful-agentic-AI-development">into AI development</a>.</p> <section class="section main-article-chapter" data-menu-title="Creating a competitive edge"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Creating a competitive edge</h2> <p>PostgreSQL, an open source format for storing disparate data types, has emerged as a popular means of managing data for agents and other AI applications. By 2024, PostgreSQL had become <a target="_blank" href="https://survey.stackoverflow.co/2024/technology" rel="noopener">the most popular database</a>, ahead of MySQL, Microsoft SQL Server, MongoDB and Redis.</p> <p>Versatility -- storing geospatial, time series, JSON and vector database workloads -- is one reason for PostgreSQL's popularity. A large, vibrant community that contributes to its evolution is another.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> The acquisition addresses a critical infrastructure gap in the emerging world of agentic applications, where AI agents need both ultra-low latency access to local data in sandboxed environments and real-time synchronization with centralized systems. </figure> <figcaption> <strong>Stephen Catanzano</strong>Analyst, Omdia </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>Given PostgreSQL's growing role in the AI pipeline, Databricks, <a href="https://www.techtarget.com/data-technologies/news/366625068/Snowflake-acquisition-of-Crunchy-Data-adds-Postgres-database">Snowflake</a> and <a href="https://www.techtarget.com/data-technologies/news/366633563/Streaming-vendor-Redpanda-buys-SQL-engine-unveils-AI-suite">Redpanda</a> all acquired PostgreSQL database capabilities in 2025. Now, by adding Electric's embeddable PostgreSQL databases, Databricks is extending such capabilities beyond traditional data and AI pipelines. In addition, Electric founders James Arthur and Kyle Matthews are joining Databricks as part of the acquisition.</p> <p>Based in Berkeley, Calif., Electric is the developer of <a href="https://pglite.dev/docs/about">PGlite</a>, a lightweight WebAssembly (WASM) version of a PostgreSQL database small enough to be embedded into agents and other applications. With PGlite embedded, applications can <a href="https://www.techtarget.com/data-technologies/news/366646252/Context-is-king-as-agents-evolve-but-problems-persist">access appropriate context</a> where they run rather than having to call back to traditional databases and other data repositories.</p> <p>In addition, Electric provides a real-time sync engine that connects data between agents and a centralized PostgreSQL architecture so that agents can access an organization's data of record in the cloud, as well as the local context that enables them to deliver outputs based on relevant data.</p> <p>Once Electric's capabilities are integrated with Lakebase, Databricks users will be able to deploy swarms of agents that stay current while working alongside one another.</p> <p>"Lakebase already made Postgres fast for agents," Devin Pratt, an analyst at IDC, told TechTarget. "Electric moves it the last few inches, right into the sandbox itself."</p> <p>Meanwhile, though Databricks competitors such as AWS, Snowflake and Microsoft provide PostgreSQL databases, the acquisition of Electric adds embeddable PostgreSQL capabilities that <a href="https://www.techtarget.com/data-technologies/news/366617350/Record-funding-round-reflects-Databricks-differentiation">distinguish Databricks</a> from its peers, Pratt continued.</p> <p>"Snowflake bought its Neon, [but] nobody's bought their Electric yet," he said. "That's the gap Databricks just closed on itself."</p> <p>Catanzano similarly noted that adding Electric's capabilities will help Databricks stand apart from competitors. While other providers offer managed PostgreSQL and data synchronization services, they are designed for traditional database workloads rather than the unique requirements of agents that access the data they need at runtime, operate in distributed sandboxed environments, and <a href="https://www.techtarget.com/data-technologies/news/366646162/Potential-consequences-are-severe-when-AI-agents-lack-context">require both individualized context</a> and a shared state.</p> <p>"Electric's combination of embeddable WASM Postgres with real-time sync architecture specifically designed for agent collaboration represents a more specialized solution for this emerging use case, positioning Databricks ahead of the curve … rather than simply keeping pace," Catanzano said.</p> <p>Leone, however, noted that while Electric's capabilities provide some differentiation, they aren't completely unique. For example, startup Turso similarly provides each agent with its own small database.</p> <p>"What's less common is Databricks insisting the small copy next to the agent be the exact same database as the big one in the middle, not a lookalike," he said.</p> </section> <section class="section main-article-chapter" data-menu-title="Additive acquisitions"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Additive acquisitions</h2> <p>While acquiring Electric adds capabilities that help distinguish Databricks among the vendors racing to add tools that help customers build AI tools, it's just the latest in a long line of acquisitions Databricks has made over the past few years to quickly add capabilities that fuel AI development.</p> <p>In addition to Neon and Electric, Databricks acquisitions include <a href="https://www.techtarget.com/data-technologies/news/366542389/Databricks-acquiring-MosaicML-to-add-more-generative-AI">MosaicML</a>, which forms the foundation of the Mosaic AI development platform, along with Arcion, BladeBridge, Einblick, Lilac AI, Mooncake and Tabular to add capabilities that complement Mosaic AI and Lakebase.</p> <p>"They've been incredibly smart, and unusually consistent about it," Leone said. "Databricks buys things it would otherwise spend years building itself, and it buys the people along with the technology, which is why the same names keep showing up. Electric fits that perfectly."</p> <p>In Electric's case, engineering a full database to run in a small workspace such as an agent's <a href="https://www.techtarget.com/cybersecurity/definition/sandbox">sandbox</a> is different than engineering a database to run in the cloud, he continued.</p> <p>"Electric has already solved it," Leone said. "The Electric team is joining the same group that came over with Neon, so that's two database companies now folded into one team on purpose."</p> <p>Pratt likewise noted that Databricks' acquisition strategy has been effective to date. In particular, MosaicML and Neon have become key parts of the vendor's platform, though Tabular's capabilities, which added support for Apache Iceberg storage, have been slower to integrate, he continued.</p> <p>"Databricks isn't buying new products, it's buying pieces that already have a home waiting for them," Pratt said.</p> </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Looking ahead</h2> <p>With Databricks <a href="https://www.techtarget.com/data-technologies/news/366636532/Databricks-adds-4B-funding-round-IPO-could-be-next">continuing to raise capital</a> -- the vendor revealed on Thursday that it added $5 billion in venture capital funding that would bring its total funding to over $30 billion -- more acquisitions are possible.</p> <p>Leone, however, advised Databricks to concentrate at least some resources on delivering new capabilities it <a href="https://www.techtarget.com/data-technologies/news/366644813/Databricks-intros-new-Genie-data-management-tools-to-aid-AI">publicly revealed in recent months,</a> but hasn't yet made generally available. In addition, making its growing platform easier to navigate would be wise, Leone continued.</p> <p>"A lot of what Databricks announced this year is still labeled beta or preview, meaning it isn't fully released," he said. "Getting that work done is worth more to current customers than another round of new features. I'd also want them to say plainly which tool to use for which job, because the platform has gotten big enough that its size is now the problem."</p> <p>Pratt, meanwhile, suggested that Databricks continue to ensure that the capabilities it is acquiring get integrated in a manner that makes them both additive as well as easy to use with the rest of <a href="https://www.techtarget.com/data-technologies/news/366639354/New-Databricks-tool-targets-streaming-data-cost-complexity">the vendor's platform</a>.</p> <p>"Neon to Lakebase took about a year -- that's the pace the rest of the portfolio should be held to," he said.</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> Flush with funding, the data management and AI vendor continues to make purchases that add capabilities aimed at aiding users building and deploying agents. https://cdn.ttgtmedia.com/rms/onlineimages/money_g1250581414.jpg https://www.techtarget.com/data-technologies/news/366649200/Databricks-Electric-acquisition-adds-embeddable-PostgreSQL Thu, 13 Aug 2026 17:15:00 GMT <p>MongoDB on Thursday unveiled new features designed to connect operational data with AI development tools so users can easily and quickly feed agents and other applications the context they need to deliver accurate outputs.</p> <p>Revealed during MongoDB.local Build Fest, a user event in San Francisco, the new capabilities, among others, include connectors and integrations with agentic coding platforms such as Grok Build and Vercel, a managed <a href="https://www.techtarget.com/data-technologies/feature/One-year-of-MCP-Support-a-must-for-data-management-vendors">Model Context Protocol (MCP) server</a> for MongoDB's Atlas database service, and automated Voyage AI <a href="https://www.techtarget.com/data-technologies/feature/Vector-search-now-a-critical-component-of-GenAI-development">vector embeddings</a> in MongoDB Atlas.</p> <p>Collectively, MongoDB's latest set of tools are significant additions for existing users given that they unify capabilities in one platform that otherwise must be pieced together with tools from different vendors, according to Stephen Catanzano, an analyst at Omdia, a division of TechTarget.</p> <p>"This is a strong set of capabilities that collectively solve … the operational complexity and data staleness that comes from stitching together separate systems for databases, vector stores and embeddings," he told TechTarget. "They unify live operational data with AI retrieval and agent workflows, eliminating synchronization overhead while letting developers work inside tools they already use."</p> <p>Unification, in fact, is how MongoDB is distinguishing itself from competing vendors that are also introducing features designed to aid AI development, Catanzano continued.</p> <p>"Competitors typically require teams to bolt together separate operational databases, vector stores and embedding services that need constant synchronization," he said. "The emphasis on real-time data access combined with Voyage AI models and automated embedding management that eliminates custom pipelines, sets it apart from the fragmented architectures most vendors require."</p> <p>Based in New York City, MongoDB is a longtime database specialist that, like competing database vendors such as <a href="https://www.techtarget.com/data-technologies/news/366645317/Couchbase-evolution-continues-with-new-data-layer-for-AI">Couchbase,</a> along with broader data management vendors such as <a href="https://www.techtarget.com/data-technologies/news/366644813/Databricks-intros-new-Genie-data-management-tools-to-aid-AI">Databricks</a> and <a href="https://www.techtarget.com/data-technologies/news/366643795/Snowflake-barrage-adds-more-AI-development-analysis-tools">Snowflake</a>, has expanded into AI development over the past few years.</p> <section class="section main-article-chapter" data-menu-title="Aiding AI development"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Aiding AI development</h2> <p>Despite the high priority data management and AI providers have placed on simplifying AI development, building agents and other AI tools that perform as intended in production remains challenging. Heading into 2026, <a target="_blank" href="https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html?id=us:2ps:3gl:aisgm26:awa:CONS:em:K0218784:012626:kwd-430833501819:195648817121:794247818306::&amp;gclsrc=aw.ds&amp;gad_source=1&amp;gad_campaignid=23269751971&amp;gclid=CjwKCAjws_DTBhB_EiwAXZknGeJC8uqNU9YNaTWHUXJEhY_qQNxuAtQrd6VPywO0mCMZiX_C4vmHshoCupUQAvD_BwE" rel="noopener">Deloitte reported</a> that only one-quarter of all enterprises expected even 40% of their AI projects to ever make it past the pilot stage.</p> <p>In response, vendors ranging from tech giants such as AWS and Microsoft to specialists including Alteryx and Tableau have all introduced features designed to enable customers to more successfully build AI tools. In particular, developing tools that enable enterprises to build a data layer that <a href="https://www.techtarget.com/data-technologies/news/366646252/Context-is-king-as-agents-evolve-but-problems-persist">feeds AI tools relevant context</a> has been a focal point for many vendors.</p> <p>That includes MongoDB.</p> <p>In January, the vendor <a href="https://www.techtarget.com/data-technologies/news/366637414/MongoDB-launches-latest-Voyage-models-to-aid-AI-development">launched new Voyage AI models</a> to improve data retrieval for agents, in May MongoDB <a href="https://www.techtarget.com/data-technologies/news/366642768/MongoDB-adds-new-vector-performance-capabilities-to-aid-AI">released new vector indexing capabilities</a> and in June <a href="https://www.techtarget.com/data-technologies/news/366645316/Latest-MongoDB-tools-tackle-top-AI-development-hurdles">introduced tools</a> that further address feeding agents with relevant context.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> This is a strong set of capabilities that collectively solve … the operational complexity and data staleness that comes from stitching together separate systems for databases, vector stores and embeddings. </figure> <figcaption> <strong>Stephen Catanzano</strong>Analyst, Omdia </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>Now, MongoDB is continuing its efforts to aid AI development more capabilities&nbsp;</p> <p>The addition of the specific capabilities unveiled on Wednesday was motivated by a mix of customer feedback and MongoDB's own observations of how the AI development landscape is evolving, according to Ben Cefalo, the vendor's chief product officer.</p> <p>"The tools people build will change quickly, and if MongoDB isn't already present wherever that work is happening, [we would be] asking builders to step outside their workflow to use us," he said. "That's the wrong direction. … For us, it's the same principle MongoDB was founded on, [which is to] meet developers on their own terms, wherever they are."</p> <p>The new features with the most potential value for users are the managed MCP server for Atlas and Automated Voyage AI vector embeddings in MongoDB Atlas, according to Catanzano.</p> <p>The first provides developers with a managed version of <a href="https://www.techtarget.com/data-technologies/news/366631455/MongoDB-adds-MCP-server-expands-AI-development-capabilities">the vendor's MCP server</a>, which was first launched in September 2025 and enables users to standardize connections between agents and MongoDB. The second automates &nbsp;the generation of vectors that make data discoverable to agents.</p> <p>"The managed MCP server provides a zero-maintenance, standards-based way for any agent to access MongoDB data with proper governance controls, while automated embeddings solve the painful synchronization problem by generating and updating embeddings inside the database as documents change," Catanzano said.</p> <p>In addition to the managed MCP server for Atlas and Automated Voyage AI vector embeddings in MongoDB Atlas, MongoDB's new capabilities include the following:</p> <ul type="disc" class="default-list"> <li>Integrations with agent-powered coding platforms Cursor, Devin, Grok Build and Vercel, including a native integration between MongoDB Atlas and <a href="https://www.nocode.mba/articles/v0-review-ai-apps">v0 by Vercel</a>.</li> <li>Native connectors between real-time operational data and applications built with Claude and ChatGPT.</li> <li>Atlas App Connections, an identity and access layer that enables users to track and audit an agent's actions.</li> <li>An API that makes MongoDB's embedding and reranking models available through Atlas and accessible to any application whether inside MongoDB's environment or not.</li> <li>A new AI model that enables coding agents to retrieve information from <a href="https://www.techtarget.com/whatis/definition/codebase-code-base">codebases</a> with higher accuracy and lower cost than when doing so from general-purpose embedding models.</li> <li>Vector search in Atlas Stream Processing.</li> </ul> <p>Most of the new capabilities are not yet generally available, although Automated Voyage AI vector embeddings in MongoDB Atlas, the Atlas Embedding and Reranking API, and vector search in Atlas Stream Processing are GA.</p> <p>"This is a strong new set of capabilities that strengthens MongoDB's position as a leading multi-modal database that supports agentic AI," Kevin Petrie, an analyst at BARC U.S., told TechTarget. "MongoDB correctly recognizes that AI projects require diverse structured and unstructured data inputs, along with diverse search and retrieval methods. Its combination of [capabilities] address these needs."</p> <p>Like Catanzano, Petrie noted the value of the managed MCP server for Atlas. In addition, he highlighted the significance of the connectors for users.</p> <p>"The native LLM connectors and MCP server help standardize integration with the rich ecosystem of AI elements," he said.</p> </section> <section class="section main-article-chapter" data-menu-title="Looking ahead"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Looking ahead</h2> <p>As MongoDB plots product development over the final months of 2026, the vendor plans to continue focusing on better enabling the developers and startup companies creating <a href="https://www.techtarget.com/ai/feature/11-real-world-agentic-AI-examples-and-use-cases">innovative AI applications</a> to build new tools, according to Cefalo.</p> <p>"That's a big part of why you're seeing us show up so heavily in San Francisco," he said.</p> <p>In addition, MongoDB aims to increase awareness of its Voyage AI models given how they can enable organizations to run elements of the <a href="https://www.techtarget.com/ai/tip/Agentic-AI-workflows-Trends-examples-and-best-practices">agentic AI workflow</a>, such as vector indexing and data retrieval, without having to piece together tools from various vendors, Cefalo continued.</p> <p>"The direction is consistent: fewer systems, better accuracy and less babysitting infrastructure," he said.</p> <p>Catanzano, meanwhile, suggested that MongoDB add more agent governance capabilities as more organizations begin to move agents into production environments. In addition, he noted that prebuilt agent templates for common uses such as customer support and <a href="https://www.techtarget.com/ai/opinion/How-to-manage-the-gap-between-enterprise-AI-use-and-AI-regulation">compliance monitoring</a> could aid customers that lack the expertise to build such tools from scratch.</p> <p>"MongoDB could deepen agent governance with comprehensive observability and policy enforcement for production deployments, build cost optimization features that intelligently route between models based on accuracy and budget tradeoffs, and offer prebuilt agent templates for common enterprise use cases," he said.</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> Unifying frequently disparate capabilities such as vector embedding, data retrieval and agent workflows helps distinguish the vendor amid a crowded field of data and AI providers. https://cdn.ttgtmedia.com/rms/onlineimages/machine%20learning_g1186820873.jpg https://www.techtarget.com/data-technologies/news/366649199/MongoDB-intros-latest-features-to-fuel-AI-development Thu, 13 Aug 2026 17:03:00 GMT <p>IT hardware has always been a critical and costly aspect of modern business, but the rise of AI has placed dynamic new demands on this infrastructure. To deal with these new requirements, a comprehensive set of metrics have been created to track the health, efficiency and cost of infrastructure resources and services.</p> <p>Proper utilization metrics help companies maximize machine learning (ML) model performance, prevent bottlenecks, ensure availability and manage the ever-increasing costs of cloud and local infrastructure. <a href="https://www.techtarget.com/searchitoperations/definition/IT-monitoring">IT monitoring</a> metrics paint a picture of the environment over time, letting business and technology leaders see trends, plan for future AI infrastructure investments and ensure cost-effective AI platforms. AI infrastructure metrics typically fall into four categories: compute, network, storage and business performance.</p> <section class="section main-article-chapter" data-menu-title="Compute metrics for AI infrastructure"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Compute metrics for AI infrastructure</h2> <p>Compute metrics are at the heart of AI infrastructure. ML and AI depend on advanced processors, including <a href="https://www.techtarget.com/whatis/feature/GPUs-vs-TPUs-vs-NPUs-Comparing-AI-hardware-options">GPUs, tensor processing units (TPUs) and neural processing units</a> for high-performance mathematical operations. Processors, memory and model use define AI performance and corresponding cost efficiency.</p> <p>A cross-section of important metrics for AI compute characteristics includes the following:</p> <ul class="default-list"> <li><b>GPU memory saturation</b> occurs when video random access memory (<a href="https://www.techtarget.com/searchstorage/definition/video-RAM">VRAM</a>) on the GPU subsystem is fully used and the GPU then must access slower system RAM across suboptimal interfaces, such as the PCIe bus. Saturation will cause severe performance degradation and result in out-of-memory errors during ML training and inference. GPU memory should generally not exceed 60%-80% utilization to ensure adequate space for model parameters, gradients and other data elements.</li> <li><b>GPU memory bandwidth</b> is the rate of data transfer between a GPU and its VRAM, which provides a high-speed workspace for storing GPU data. The metric is measured in gigabytes per second (GB/s) or terabytes per second. The critical issue is to ensure enough bandwidth where VRAM data can flow without disruption. If data can't feed the GPUs fast enough, it pauses while it waits for the VRAM to catch up. Inadequate GPU memory bandwidth will slow ML training and inference.</li> <li><b>GPU memory capacity </b>is the total amount of memory available to the GPUs measured in GBs. Capacity is typically coupled with saturation metrics to gauge overall GPU memory use. If utilization regularly reaches saturation, additional memory capacity might be justified. In 2026, GPU memory capacity varies from 16 GB to 140 GB per GPU depending on the workload and AI infrastructure.</li> <li><b>GPU utilization</b> is the percentage of time that a GPU or other processor accelerator technology is active. The goal is to see moderately high levels of use. Typical GPU utilization characteristics include 80% to 95% for training cycles, and 70% to 90% for ongoing inference. Lower percentages suggest that the GPU is underutilized, reducing the cost-effectiveness of the <a href="https://www.techtarget.com/searchenterpriseai/tip/How-to-choose-the-best-GPUs-for-AI-projects">GPU investment</a>. Higher percentages mean that the GPU might be overutilized and could become a bottleneck to AI performance, so additional capacity might be justified.</li> <li><b>Hardware memory error rates</b>, such as bit flips, are common and almost all are readily correctable on the fly. However, AI creates enormous stress on high-performance memory subsystems and interconnects, and heavy data transfer loads can often drive higher hardware memory errors. Each error delays data transfers and processing while it's corrected or, if left uncorrected, while it reprocesses. Tracking memory error rates and throughput metrics can help infrastructure experts optimize compute performance, identify unreliable memory subsystems and can justify the investment in high-quality memory using advanced error correction techniques.</li> <li><b>ML model latency</b> is the amount of time for a model to process a request and return a response. It's typically measured in milliseconds per request. Throughput and latency are often used together as an overall measure of <a href="https://www.techtarget.com/searchenterpriseai/feature/Optimize-AI-models-to-generate-more-bang-for-your-buck">model efficiency</a>. For example, high throughput and low latency suggest that the model is operating at high efficiency.</li> <li><b>ML model throughput </b>provides a measure of the model's performance by determining the number of requests a model can handle per second. It can be measured in many places. Falling throughput suggests the presence of bottlenecks, unoptimized or poorly configured infrastructure.</li> <li><b>Model floating-point operations per second (FLOPS) utilization</b> is a popular standard for training efficiency. MFU is the ratio of achieved theoretical compute performance, measured in billions of FLOPS or TFLOPs to peak hardware capacity. Ideally, the model achieves full utilization of hardware capacity. In actual practice, however, an MFU of 40% or better typically indicates a healthy, well-performing model.</li> </ul> </section> <section class="section main-article-chapter" data-menu-title="Network metrics for AI infrastructure"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Network metrics for AI infrastructure</h2> <p>Although GPUs and other processing technologies take the AI spotlight, computing prowess is worthless without the ability to move data. Networking provides the vital interconnectivity needed to move data to and from GPUs, storage and systems. An AI infrastructure must support high-throughput, low-latency data pipelines, such as 400 Gigabit Ethernet, 800 GbE or InfiniBand, and seamless interGPU communication.</p> <p>Several metrics can provide useful insights into <a href="https://www.techtarget.com/searchcloudcomputing/tip/How-to-optimize-networks-for-AI-workloads-in-the-cloud">network performance</a>, such as the following.</p> <ul class="default-list"> <li><b>InterGPU bandwidth or throughput</b> represents the speed of data transfers between GPUs or other processors within a single server. The goal is to ensure that there's enough bandwidth to keep compute processors busy. Insufficient bandwidth will starve processors of data and leave them idle, wasting time and capital investment.</li> <li><b>InterGPU latency</b> represents the time needed for a data packet to travel between GPUs. Low latency typically indicates there are no bottlenecks. It's normal for latency to increase as AI systems become busier, such as during training and inference. However, excessive latency suggests bottlenecks or an unoptimized infrastructure design.</li> <li><b>Internode latency </b>gauges the time needed for data to travel between AI nodes or servers often fitted with high-performance network links such as InfiniBand or remote direct memory access (<a href="https://www.techtarget.com/searchstorage/definition/Remote-Direct-Memory-Access">RDMA</a>) over converged ethernet or RoCE. As with interGPU latency, low latency and high throughput internode performance is essential for maintaining synchronization in distributed ML training or AI inference processes.</li> <li><b>Queue depths and buffer use</b> are important because networks rely on queues or buffers to hold data briefly while waiting for network availability. A well-designed AI fabric uses networking architectures intended for high bandwidth and low latency, which should keep queue or buffer use to a minimum. Increasing queue or buffer use suggests network congestion, which stalls data movement and leaves processors idle.</li> <li><b>AI fabric packet loss</b> is another important networking metric. Some lost or dropped packets are normal in networking, and lost data can simply be resent. However, the time to resend lost packets can impair ML training and AI inference performance. <a href="https://www.techtarget.com/searchdatamanagement/tip/AI-data-fabric-emerges-as-a-governance-layer-for-agents">AI fabrics</a> depend on high-bandwidth, low-latency, zero-loss networks to maintain optimum MFU. This requires careful and dedicated design.</li> <li><b>RDMA or RoCE drops </b>is a metric that measures errors or losses when advanced server technologies, such as RDMA and RoCE, directly exchange memory data between systems without going through a CPU. This approach vastly improves data exchange performance and helps infrastructure experts identify and mitigate GPU synchronization problems<b>.</b> CPUs typically handle data movement to and from memory.</li> </ul> </section> <section class="section main-article-chapter" data-menu-title="Storage metrics for AI infrastructure"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Storage metrics for AI infrastructure</h2> <p>Effective AI is dependent on enormous volumes of data, often stored in data warehouses or vast data lakes. This data must be exchanged with servers across a network, but <a href="https://www.techtarget.com/searchstorage/tip/Enterprises-Face-New-Storage-Bottlenecks-as-AI-Grows">storage resources play a part in an effective AI infrastructure</a>.</p> <p>The following common storage metrics can be extended to support complex AI infrastructures.</p> <ul class="default-list"> <li><b>Storage throughput</b> measures the total volume of data moved over time. It's a useful metric because storage subsystems must provide massive blocks of data to keep GPUs and AI processing pipelines filled. Read/write throughput can be measured in GB/s. Falling or disrupted figures suggest storage bottlenecks that should be investigated and remediated.</li> <li><b>Storage IOPS </b>measures the quantity of individual storage I/O operations over time and is a close corollary to storage throughput. While both metrics are important for storage, <a href="https://www.techtarget.com/searchstorage/definition/IOPS-input-output-operations-per-second">IOPS</a> can be particularly meaningful for small batch files, random inference queries and metadata-centric workloads. IOPS disruptions indicate possible bottlenecks in the storage subsystem.</li> <li><b>Storage latency</b> is the time needed to transfer data from storage to GPU memory for processing. Ideally, storage subsystems are selected and networked to provide very low latency. Increasing storage latency can reduce model training efficiency by slowing weight updates or delaying the transfer of data collected from IoT and other real-world sources.</li> <li><b>Storage capacity and utilization</b> metrics monitor the total amount of storage in the environment as well as the amount of storage used. They are both often well into the terabyte range and sometimes into the petabyte range. As data volumes and retention demands increase, free storage capacity typically declines over time, providing an objective justification for additional storage upgrades.</li> </ul> </section> <section class="section main-article-chapter" data-menu-title="Tools for AI hardware metrics"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Tools for AI hardware metrics</h2> <p>Metrics require tools for proper data collection and reporting. There are many tools available, each with unique strengths and weaknesses, as well as some specialized ones for specific tasks. Some tools focus on infrastructure for tracking the behavior of core components and systems, such as processors, memory and networks. AI observability platforms <a target="_blank" href="https://www.gartner.com/reviews/market/ai-evaluation-and-observability-platforms" rel="noopener">correlate hardware metrics with model performance</a>, tracing and token use. Model profilers examine hardware performance and identify hardware AI training bottlenecks.</p> <p>The following list provides a sampling of tools gathered from vendor websites and other current sources:</p> <ul class="default-list"> <li><b>Coralogix.</b> An observability platform that provides logs, metrics, traces and security data and analysis.</li> <li><b>Datadog.</b> An AI observability and security platform that unifies metrics, logs, traces and monitors to troubleshoot performance issues, optimize cloud infrastructure and secure systems.</li> <li><b>Grafana Cloud.</b> A fully managed AI observability platform for collecting and unifying metrics, logs, traces, profiles, dashboards and other infrastructure data.</li> <li><b>LogicMonitor.</b> A managed AI observability platform with visibility into complex IT infrastructures including cloud, local, and edge environments.</li> <li><b>MLPerf Storage.</b> A tool that provides benchmarks to characterize the performance of storage subsystems supporting ML workloads.</li> <li><b>New Relic AI Observability.</b> An end-to-end monitoring platform that correlates model behavior to traditional system metrics and providing visibility into LLMs, vector databases and agentic frameworks.</li> <li><b>NVIDIA Base Command Manager.</b> Cluster management software that automates provisioning, monitoring and management of high-performance computing and AI server clusters.</li> <li><b>NVIDIA DCGM Exporter</b>. A tool to gather GPU metrics and observe workload behavior or monitor GPU clusters.</li> <li><b>Prometheus.</b> An open source systems monitoring toolkit for cloud-native environments to collect and report metrics from applications and infrastructure.</li> <li><b>PyTorch Profiler.</b> A PyTorch feature that collects performance metrics, such as execution time, GPU kernel utilization and memory consumption during training and inference.</li> <li><b>TensorFlow Profiler.</b> Tools included with TensorBoard that measure execution performance and resource consumption of ML models, identify bottlenecks across hardware resources, and optimize training speed and memory efficiency.</li> <li><b>Zabbix.</b> An open source IT monitoring platform to track the performance, availability and health of servers, virtual machines, networks, cloud services and applications.</li> </ul> <p><i>Stephen J. Bigelow, senior technology editor at TechTarget, has more than 30 years of technical writing experience in the PC and technology industry.</i></p> </section> AI infrastructure requires monitoring compute, network, storage and business metrics to optimize performance, prevent bottlenecks and manage costs. https://cdn.ttgtmedia.com/rms/onlineimages/ai_a373894778.jpg https://www.techtarget.com/searchenterpriseai/tip/Essential-metrics-for-AI-hardware-management Thu, 30 Jul 2026 23:39:00 GMT <p>AI is redefining how businesses evaluate data center providers.</p> <p>Traditional procurement criteria, such as uptime, cost and basic reliability, no longer reflect the realities of GPU-intensive AI training and inference. <a href="https://www.techtarget.com/searchcloudcomputing/tip/Is-your-compute-strategy-ready-for-AI-workloads-in-the-cloud">AI workloads demand extreme power density</a>, sustained utilization, purpose-built cooling and low-latency interconnects. Most legacy data centers were never designed for these demands.</p> <p>AI data centers operate at a scale comparable to energy or industrial infrastructure builds. These projects use more power and water, and have larger physical footprints than traditional data centers. Supplying enough power, cooling and space are key design challenges rather than secondary considerations when building one of these facilities. In addition, AI training environments depend on large GPU clusters that exchange massive volumes of data in real time, making static capacity planning increasingly impractical.</p> <p>"Many facilities might support current AI inference workloads, but newer reasoning models, agentic systems and hyperscale deployments are driving GPUs toward longer runtimes and higher utilization," said Adam Morton, CTO of data center business at Flex, a company that designs, builds and deploys AI infrastructure systems.</p> <p>Faced with these constraints, companies selecting a data center provider must assess a range of factors. "Buyers need to evaluate not only whether a facility can support today's workloads, but whether its infrastructure can adapt as AI requirements continue to change," Morton said.</p> <p>AI data center projects also face increased public scrutiny, regulatory review and community pushback. Whether they proceed in a timely manner often depends on local and state permitting timelines, utility constraints, and <a href="https://www.techtarget.com/searchenterpriseai/feature/Communities-call-for-transparency-in-AI-data-center-deals">local lobbying and legal challenges</a> over land use, power grid, water, noise and other environmental concerns. In 2025 alone, local opposition contributed to delays or cancellations of projects totaling $156 billion in planned investment, according to a Data Center Watch <a href="https://www.datacenterwatch.org/q3-q4-2025">report</a>.</p> <p>In this new data center landscape, businesses looking to hire an AI data center provider must evaluate whether that provider has the technical capabilities to handle increasingly demanding AI workloads. Businesses must also pay attention to and ask questions about the conversations going on in the communities and states where the facilities are located: What are the concerns, how could they affect the provider's services and how is the provider responding?</p> <section class="section main-article-chapter" data-menu-title="The new reality of AI data center selection"> <h2 class="section-title"><i class="icon" data-icon="1"></i>The new reality of AI data center selection</h2> <p><a href="https://www.techtarget.com/searchenterpriseai/tip/How-to-choose-a-data-center-for-AI-workloads">Picking an AI data center provider</a> is no longer mostly a technical procurement decision with sustainability considerations tacked on later. These decisions now require risk assessment that spans GPU performance, power and cooling capacity, environmental impact, regulatory pressures and community opposition.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> The most expensive GPU in the world creates no value while waiting for the rest of the system to catch up. </figure> <figcaption> <strong>Satyam Dhar</strong>Software engineer at Galileo </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>At the same time, AI-ready marketing claims are outpacing the infrastructure behind them. Providers might present planned capacity as if it already exists, while downplaying environmental concerns and local opposition. Assessing these providers means moving beyond broad claims to determine if the infrastructure is already operational, what still depends on future approvals or upgrades, and where long-term risks exist.</p> <p>Businesses that ask the right questions early protect more than just technology investments. They're also reducing the risk of future expansion delays, operational disruptions and reputational issues associated with controversial infrastructure projects. Ultimately, the goal isn't simply to secure access to GPUs but to ensure the surrounding infrastructure is available when needed and can support those needs efficiently and at scale.</p> <p>A clear sign of infrastructure problems is idle hardware rather than system failures, said Satyam Dhar, a software engineer at Galileo, an AI evaluation and observability platform company owned by Cisco. "The most expensive GPU in the world creates no value while waiting for the rest of the system to catch up."</p> <p>In many cases, the success of an AI strategy depends as much on having the right information about a data center's power, cooling, networking, planned infrastructure and provider transparency as on the AI technology itself.</p> </section> <section class="section main-article-chapter" data-menu-title="Technical questions to ask data center vendors"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Technical questions to ask data center vendors</h2> <p>Technical qualification is the first step when establishing whether a facility can physically support AI workloads at the required scale. Technical questions to ask include the following:</p> <h3>1. Power density and capacity planning</h3> <p>Power is often the biggest factor in determining whether AI infrastructure can successfully scale. One of the first questions enterprise executives should ask data center vendors is how much rack power density its facilities support and whether it aligns with their company's current and future AI workload requirements.</p> <p>Other important considerations include how additional capacity will be provided for over time and whether expansion depends on utility approvals, transmission upgrades or infrastructure projects still in development.</p> <p>"Providers should be clear about what's secured and what's still planned, rather than presenting future capacity as existing capacity," said Sam V. Tabar, CEO of WhiteFiber, an AI infrastructure provider that offers GPU cloud services and AI-focused data center capacity.</p> <p>Businesses should also examine whether the facility's energy mix relies on renewables, natural gas, coal, nuclear or a combination of sources. The mix can affect both operational resilience and sustainability goals. For example, facilities that rely on a diverse set of energy sources or have backup generation strategies might be better positioned to maintain operations during grid disruptions or periods of energy constraints.</p> <p>Sustainability claims should be backed by independent verification; that's a standard that applies across all vendor conversations, not just power-specific ones.</p> <div class="extra-info"> <div class="extra-info-inner"> <h3 class="splash-heading">Key power density and capacity planning questions to ask</h3> <ul type="disc" class="default-list"> <li>What rack power density does the facility support?</li> <li>How is future power expansion expected to be generated?</li> <li>What energy sources power the facility and can they support future growth?</li> <li>Are sustainability and emissions claims independently verified?</li> </ul> </div> </div> <h3>2. Cooling architecture</h3> <p>Traditional air-cooling systems were designed for lower-density CPU workloads and aren't sufficient for modern GPU deployments. As AI rack densities increase, operators are turning to liquid cooling to sustain performance and manage heat.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> The power side of the data center has decades of layered defense in depth. The cooling side often has one layer. </figure> <figcaption> <strong>Adam Morton</strong>CTO of Flex </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>Liquid cooling is no longer a premium feature, WhiteFiber's Tabar said. "Liquid cooling should be viewed as a baseline requirement, not an upgrade option for modern GPU clusters," he said.</p> <p>Businesses should look beyond whether liquid cooling exists and evaluate how resilient the cooling architecture remains under sustained load, Flex's Morton said. The greatest operational risk in many AI environments isn't the power infrastructure but the cooling infrastructure, especially the coolant distribution systems and secondary fluid networks that serve multiple racks simultaneously.</p> <p>Electrical systems are typically built with multiple layers of redundancy, but cooling systems often have fewer backup mechanisms, Morton explained. "The power side of the data center has decades of layered defense in depth," he said. "The cooling side often has one layer."</p> <p>Power failures are often localized and recoverable, Morton added, whereas cooling failures can affect multiple racks and GPU clusters simultaneously, potentially increasing downtime and recovery times. Organizations should evaluate how providers design redundancy, maintenance protocols and failure response scenarios into their cooling systems.</p> <p>Beyond system reliability, resource inputs are also becoming a consideration. Facilities that rely heavily on freshwater for cooling might face greater exposure to drought conditions, water restrictions and regulatory scrutiny, making water use an important factor in site selection.</p> <div class="extra-info"> <div class="extra-info-inner"> <h3 class="splash-heading">Key cooling infrastructure questions to ask</h3> <ul type="disc" class="default-list"> <li>What cooling systems are used?</li> <li>Can the cooling system support higher-density AI workloads over time?</li> <li>How is cooling redundancy designed, and what happens if a cooling system fails?</li> <li>Does the cooling infrastructure rely heavily on freshwater sources?</li> <li>Could water use create future regulatory or community challenges?</li> </ul> </div> </div> <h3>3. Network and interconnect performance</h3> <p>AI training workloads are far more sensitive to network delays than traditional enterprise applications. Businesses should evaluate how the data center manages latency in distributed AI environments, what high-bandwidth networking options are available and how communication among GPU nodes is optimized.</p> <p>Such limitations usually become most noticeable when organizations scale up training across large GPU clusters, Galileo's Dhar said. During this stage, networking bottlenecks can lead to slower training cycles, uneven data processing speeds or situations where adding more GPUs no longer delivers expected performance improvements, he said.</p> <p>Businesses using <a href="https://www.techtarget.com/searchcloudcomputing/feature/Multi-cloud-vs-hybrid-cloud-and-how-to-know-the-difference">hybrid or multi-cloud environments</a> must also assess the availability and performance of cloud connectivity options that enable fast, reliable data movement across platforms.</p> <p>Many of the networking and latency challenges within individual AI data centers are being addressed through AI-specific networking architectures, Morton said. The bigger challenge now is shifting to coordinate across multiple facilities as businesses scale beyond single-site deployments.</p> <p>"The AI factory of the next decade isn't a single building; it's a fabric of buildings," Morton said.</p> <p>As a result, the focus is shifting from individual rack- or facility-level performance toward intercampus connectivity and distributed architectures capable of supporting large-scale AI training environments.</p> <div class="extra-info"> <div class="extra-info-inner"> <h3 class="splash-heading">Key networking questions to ask</h3> <ul type="disc" class="default-list"> <li>How does the facility manage latency in distributed AI environments?</li> <li>What high-bandwidth networking and interconnect options are available?</li> <li>How is communication among GPU nodes optimized?</li> <li>What cloud connectivity options are available for hybrid or multi-cloud environments?</li> <li>How does the provider support connectivity and workload coordination across multiple facilities?</li> </ul> </div> </div> <h3>4. Scalability and deployment risk</h3> <p>Businesses should assess how fast a provider can bring on additional capacity online and what factors could delay expansion. Future growth often depends on permits, zoning decisions, utility approvals and infrastructure upgrades that are still in progress. When these factors are in play, capacity projections should be treated as contingent rather than guaranteed.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> The industry still designs every data center as a custom project. </figure> <figcaption> <strong>Adam Morton</strong>CTO of Flex </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>These dependencies aren't unusual, but rather they reflect the long timelines required for grid expansion, including transmission upgrades and transformer availability, according to Brad Johnson, director of electric utilities at Bentley Systems, an infrastructure engineering software company.</p> <p>Permit complexity is increasing as AI infrastructure evolves faster than existing regulatory frameworks, particularly for emerging technologies, such as high-density cooling systems and alternative power architectures, Morton said. External infrastructure constraints are only part of the challenge, and many deployment delays stem from a more structural issue in how data centers are designed.</p> <p>"The industry still designs every data center as a custom project," Morton said. This approach requires each facility to be individually engineered, built and commissioned, which limits the ability to scale capacity quickly even when demand is strong, he added.</p> <p>As a result, business leaders evaluating providers should look at whether deployments use standardized, repeatable designs or highly customized builds. More standardized infrastructure can reduce risk and make timelines more predictable.</p> <div class="extra-info"> <div class="extra-info-inner"> <h3 class="splash-heading">Key scalability and deployment questions to ask</h3> <ul type="disc" class="default-list"> <li>How quickly can additional AI capacity be brought online?</li> <li>Have regulatory or permit constraints affected past expansion timelines?</li> <li>How much future capacity is already secured versus still planned?</li> <li>Does expansion depend on pending permits, utility approvals or infrastructure upgrades?</li> <li>What contingency plans exist if key expansion projects are delayed?</li> <li>Are new deployments based on standardized, repeatable designs or highly customized builds?</li> </ul> </div> </div> </section> <section class="section main-article-chapter" data-menu-title="Sustainability questions to ask data center vendors"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Sustainability questions to ask data center vendors</h2> <p>Power, cooling and networking tend to dominate AI infrastructure discussions. However, some of the most material long-term risks are less visible at the procurement stage. Energy availability, water access and environmental constraints can directly affect a facility's ability to scale.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Many of the infrastructure and sustainability constraints … are often not analyzed holistically by enterprise buyers. </figure> <figcaption> <strong>Brad Johnson</strong>Director of electric utilities at Bentley Systems </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>These risks aren't hidden, but they often require looking beyond a vendor's marketing materials, Bentley System's Johnson said. "Many of the infrastructure and sustainability constraints are visible in public utility filings and regulatory processes but are often not analyzed holistically by enterprise buyers," he said.</p> <p>Key infrastructure and sustainability considerations for enterprises evaluating AI data center vendors include the following:</p> <h3>1. Utility and power stability</h3> <p>Basic power redundancy is a baseline expectation for data centers supporting AI workloads. The more important question is whether the surrounding grid can keep up with rising AI-driven energy demand.</p> <p>"Many of today's bottlenecks stem from grid physics, transmission limits, transformer availability, material shortages and engineering complexity," Johnson said.</p> <p>Energy availability has already become one of the most significant constraints on AI infrastructure growth, Morton said. AI operators are increasingly exploring alternatives, such as behind-the-meter generation, fuel cells and other forms of dedicated power infrastructure to reduce dependence on utility timelines and improve deployment certainty.</p> <p>For enterprises, this means evaluating not only a provider's current power capacity but also how future growth will be supported if regional grid expansion fails to keep pace with AI demand.</p> <p>It's also worth examining how a provider's energy mix aligns with <a href="https://www.techtarget.com/sustainability/feature/Business-sustainability-trends">enterprise sustainability</a> goals. Facilities that provide independently verified energy and emissions data typically offer greater transparency and long-term reliability than those relying mainly on renewable energy credits while still operating on fossil-fuel-heavy grids.</p> <div class="extra-info"> <div class="extra-info-inner"> <h3 class="splash-heading">Key utility and power stability questions to ask</h3> <ul type="disc" class="default-list"> <li>Can the regional power grid support long-term AI energy demand?</li> <li>Has the provider considered or implemented alternative power sources?</li> <li>How much of the facility's power is dependent on the local utility grid versus dedicated infrastructure?</li> <li>What is the facility's energy mix and how is it expected to change as demand grows?</li> <li>How does the provider secure additional power capacity if grid supply becomes constrained?</li> </ul> </div> </div> <h3>2. Water availability and cooling sustainability</h3> <p>Water consumption is becoming one of the most important and often overlooked risks in AI infrastructure. A typical data center can use an enormous amount of water per day for cooling, exposing it to drought restrictions, regulatory scrutiny and community opposition.</p> <p>At scale, the numbers are significant: Enterprise data centers consume an <a target="_blank" href="https://natureforward.org/data-centers-and-water-use/" rel="noopener">average</a> of 300,000 to 500,000 gallons of water per day, while large hyperscale facilities can draw between 1 million and 5 million gallons, comparable to the water needs of a small town. A 2025 International Energy Agency <a target="_blank" href="https://iea.blob.core.windows.net/assets/de9dea13-b07d-42c5-a398-d1b3ae17d866/EnergyandAI.pdf" rel="noopener">study</a> found that a typical 100 MW U.S. data center can require up to 2 million liters of water per day when accounting for both on-site cooling and electricity generation.</p> <p>Looking ahead, MSCI's <a target="_blank" href="https://www.msci.com/research-and-insights/blog-post/when-ai-meets-water-scarcity-data-centers-in-a-thirsty-world" rel="noopener">analysis</a> of roughly 14,000 global data center sites projected that about one in four could face increasing water scarcity risks by 2050.</p> <p>As a result, the <a href="https://www.techtarget.com/searchdatacenter/tip/Data-center-cooling-systems-and-technologies-and-how-they-work">design of cooling systems</a> is increasingly important. Liquid cooling, especially in closed-loop configurations, can significantly reduce water consumption compared with more traditional cooling approaches while also supporting higher-density AI workloads, WhiteFiber's Tabar noted.</p> <p>Businesses should evaluate how dependent a facility is on local water supplies, how much water its cooling systems use at full capacity and whether drought conditions or water restrictions could affect long-term operations or trigger regulatory scrutiny.</p> <div class="extra-info"> <div class="extra-info-inner"> <h3 class="splash-heading">Key water and cooling questions to ask</h3> <ul type="disc" class="default-list"> <li>How dependent is the facility on local water resources?</li> <li>How much water do cooling systems consume at full scale?</li> <li>Could drought conditions or water restrictions affect operations?</li> <li>Has water use already created regulatory or community concerns?</li> <li>Does the facility use closed-loop or water-reduction cooling technologies?</li> </ul> </div> </div> <h3>3. Environmental constraints and the transparency gap</h3> <p>An initial step in the evaluation process is to assess whether a formal environmental impact assessment has been completed for the existing facility and planned expansions. While the absence of one isn't necessarily a red flag, it can limit visibility into longer-term environmental risks, including emissions and water- and land-use constraints that could affect future operations.</p> <p>Transparency is another key issue. Businesses should check whether sustainability and emissions data have been independently verified, as standards vary across providers.</p> <p>These transparency challenges extend beyond individual data center providers to the broader AI ecosystem. Many tech companies emphasize <a href="https://www.techtarget.com/searchdatacenter/tip/Navigating-energy-management-strategies-in-AI-data-centers">renewable energy credits</a> and clean energy investments, even as overall emissions continue to rise.</p> <p>A 2025 <a href="https://www.itu.int/en/ITU-D/Environment/Documents/Publications/2025/Greening%20Digital%20Companies%202025%20Final.pdf">report</a> by the International Telecommunication Union found that indirect emissions from major AI-focused technology companies, including Amazon, Microsoft, Alphabet and Meta, increased by an average of 150% between 2020 and 2023, driven by rapid AI infrastructure expansion. The issue highlights a broader transparency challenge: Businesses need clearer ways to evaluate the full environmental impact of AI systems, including the infrastructure required to operate them.</p> <p>That gap is often reflected in how providers frame their own environmental performance. The issue isn't measurement itself, WhiteFiber's Tabar said, but how it's communicated externally. "What matters is that communities understand the actual resource footprint of these projects rather than relying on broad sustainability messaging," he said.</p> <p>It's also important to consider how emerging environmental regulations could affect costs and compliance requirements. New disclosure rules on energy and water use are being introduced in many regions.</p> <div class="extra-info"> <div class="extra-info-inner"> <h3 class="splash-heading">Key environmental and transparency questions to ask</h3> <ul type="disc" class="default-list"> <li>Has an environmental impact assessment been completed?</li> <li>Could future environmental regulations affect operations or expansion?</li> <li>Are sustainability and emissions claims independently verified?</li> <li>How could future disclosure requirements affect long-term costs or compliance?</li> </ul> </div> </div> <h3>4. Long-term infrastructure durability</h3> <p>Businesses should evaluate whether a facility can realistically support AI demand over the next five to ten years. That means understanding how much of the provider's future capacity plans depend on infrastructure projects that aren't yet in place.</p> <p>Enterprises should distinguish between secured capacity and announced capacity, Morton said. "A large portion of announced AI infrastructure capacity remains dependent on permits, utility approvals and transmission upgrades," he said.</p> <p>The key issues are whether providers can identify which capacity is already available and which remains contingent on external approvals, and how they would respond if critical infrastructure projects are delayed.</p> <div class="extra-info"> <div class="extra-info-inner"> <h3 class="splash-heading">Key long-term durability questions to ask</h3> <ul type="disc" class="default-list"> <li>Can the facility realistically support AI demand over the next five to ten years?</li> <li>How much future capacity depends on unfinished infrastructure projects?</li> <li>What happens if grid, transmission or water projects are delayed?</li> <li>How has the provider stress tested its long-term infrastructure plans?</li> </ul> </div> </div> </section> <section class="section main-article-chapter" data-menu-title="Questions to ask about political and community risks"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Questions to ask about political and community risks</h2> <p>AI data center expansion is increasingly becoming a public policy issue. Some local governments have introduced stricter zoning rules, development moratoriums and longer environmental review processes that can significantly delay projects and cause financial disruptions.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Operators who treat [opposition] as a permitting problem, rather than a legitimate community concern, tend to make it worse. </figure> <figcaption> <strong>Brad Johnson</strong>Director of electric utilities at Bentley Systems </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>For enterprises, these issues are no longer abstract policy debates; they're operational and reputational risks. Businesses that align their AI strategies with specific infrastructure providers can become indirectly exposed to the controversies surrounding those projects.</p> <p>Companies looking for a data center provider must discern if a facility or planned development has faced community opposition related to land use, power consumption, water use, noise or other operational effects, and how those concerns were or are being handled. A provider's response to these challenges can offer insight into its operational maturity and ability to sustain long-term expansion.</p> <p>"Operators who treat [opposition] as a permitting problem, rather than a legitimate community concern, tend to make it worse," Bentley System's Johnson said. "Many concerns are legitimate, and the industry is going to have to find solutions. Transparency and education are critical to helping projects move smoothly."</p> <p>Businesses should also evaluate whether a provider's planned expansions depend on unresolved permitting decisions, regulatory approvals or local infrastructure reviews. Delays in these processes can affect project timelines and reveal whether providers have effective strategies for managing community and government relationships.</p> <p>Businesses should ask how providers respond when community concerns, permitting challenges or regulatory reviews delay planned expansions. A provider's ability to communicate risks, engage stakeholders and adjust its plans can indicate how prepared it is to support long-term infrastructure growth.</p> <p>It's also worth looking at whether past disputes or local pushback have affected permitting timelines, expansion plans or relationships with utilities.</p> <div class="extra-info"> <div class="extra-info-inner"> <h3 class="splash-heading">Key political and community-based questions to ask</h3> <ul type="disc" class="default-list"> <li>Has the provider faced community opposition or legal challenges?</li> <li>What issues triggered public concern or local disputes?</li> <li>How has the provider addressed concerns and disputes?</li> <li>Have controversies affected permitting timelines, expansion plans or utility relationships?</li> <li>Is future expansion dependent on unresolved permits, regulatory approvals or local infrastructure reviews?</li> <li>What are the contingency plans if community or regulatory challenges delay expansion?</li> </ul> <ul class="default-list"></ul> </div> </div> <p><i>Kinza Yasar is a technical writer for Informa TechTarget's AI and Emerging Tech group and has a background in computer networking.</i></p> </section> AI growth exposes data centers to power limits, cooling issues, water scarcity and regulatory pressure. Explore the key questions to raise with vendors. https://cdn.ttgtmedia.com/rms/onlineimages/storage_g1197646065.jpg https://www.techtarget.com/ai/feature/Questions-to-ask-when-evaluating-AI-ready-data-center-providers Thu, 30 Jul 2026 11:16:00 GMT <p>As sustainability initiatives become increasingly data-driven, AI can automate labor-intensive processes, improve forecasting and uncover efficiencies that would be difficult to detect through traditional analytics.</p> <p>However, for senior IT leaders, AI presents a paradox. The same technology that can help organizations reduce waste, optimize energy consumption and improve environmental, social and governance (ESG) reporting also requires significant computing resources. <a href="https://www.techtarget.com/sustainability/tip/The-environmental-impact-of-LLMs-vs-SLMs">Large language models (LLMs) consume substantial amounts of electricity and water</a>, while the rapid expansion of AI infrastructure places new demands on power grids worldwide.</p> <p>The challenge is no longer whether AI belongs in sustainability programs, but how organizations can deploy it responsibly. Enterprise leaders must evaluate not only what AI can accomplish, but also whether its environmental benefits outweigh its own operational footprint.</p> <p>The most successful organizations approach AI as a practical optimization tool rather than a silver bullet. Instead of pursuing AI for its own sake, they apply it to measurable operational problems where they can quantify improvements in energy savings, reduced emissions or lower resource consumption.</p> <section class="section main-article-chapter" data-menu-title="What can AI do for sustainability teams?"> <h2 class="section-title"><i class="icon" data-icon="1"></i><strong>What can AI do for sustainability teams?</strong></h2> <p><a href="https://www.techtarget.com/searchcio/feature/How-CIOs-and-IT-can-drive-environmental-sustainability">Enterprise sustainability programs</a> generate enormous amounts of operational, environmental and supply chain data. AI can help organizations transform that information into actionable insights across several core areas.</p> <h3><strong>Energy grid optimization</strong></h3> <p>Machine learning (ML) models can forecast electricity demand, optimize energy distribution and improve building and data center efficiency. AI can also continuously <a href="https://www.techtarget.com/searchdatacenter/tip/Scaling-AI-data-center-cooling-for-high-density-servers">adjust HVAC systems and cooling infrastructure</a> based on changing environmental conditions and equipment loads.</p> <p>AI creates the greatest value when it is applied to physical systems with measurable outcomes, according to Jackie Swanson, managing partner at Gartner.</p> <p>"The common thread is a measurable physical output [such as] kilowatt-hours, miles [or] downtime -- data clean enough to trust and a decision someone will act on," she said. "Anything missing those three, and the value tends to live in the demo, not the P&amp;L."</p> <h3><strong>Predictive maintenance</strong></h3> <p><a href="https://www.techtarget.com/searcherp/tip/5-benefits-of-predictive-maintenance">Predictive maintenance</a> remains one of AI's most mature sustainability use cases. By analyzing telemetry data from industrial equipment, ML models can detect anomalies before failures occur, extending asset life while reducing waste, emergency repairs and unnecessary replacement.</p> <p>It also offers both financial and environmental benefits because organizations replace equipment only when necessary, rather than after catastrophic failures, said Stanislav Kazanov, head of GRC, cybersecurity, sustainability and data at Innowise.</p> <p>"Businesses save costs associated with waste, transport emissions and manufacturing overhead related to having to replace equipment on an emergency basis," he said.</p> <h3><strong>Supply chain and resource optimization</strong></h3> <p>AI also helps organizations <a href="https://www.techtarget.com/searcherp/tip/Generative-AI-use-cases-in-supply-chain">improve logistics, inventory management and demand forecasting</a>. Better forecasting reduces overproduction, minimizes waste and lets companies optimize transportation routes and procurement strategies.</p> <p>Logistics optimization and demand forecasting offer measurable results because organizations can directly quantify reductions in fuel consumption, mileage and excess inventory, Swanson said.</p> <h3><strong>Carbon footprint tracking and ESG reporting</strong></h3> <p>Many organizations use AI to automate the collection and classification of sustainability data from suppliers, invoices, shipping records and procurement systems. Natural language processing tools can significantly reduce the manual effort required to gather Scope 3 emissions data across complex supply chains.</p> <p>However, both Swanson and Kazanov cautioned against overestimating <a href="https://www.techtarget.com/sustainability/feature/How-AI-can-strengthen-ESG-reporting">AI's capabilities for ESG reporting</a>. Many AI-powered ESG platforms rely on inconsistent or incomplete data, Swanson said.</p> <p>"Point a model at data you don't trust, and you get confident, beautifully formatted estimates," she said.</p> <p>While AI can accelerate data collection, organizations should not rely on generative AI (GenAI) to create sustainability disclosures, Kazanov said.</p> <p>"A spreadsheet that has been validated by a compliance officer will always be more reliable than a summary created from an energy-consuming LLM-based system," he said.</p> </section> <section class="section main-article-chapter" data-menu-title="AI tools for sustainability"> <h2 class="section-title"><i class="icon" data-icon="1"></i><strong>AI tools for sustainability</strong></h2> <p>The market for AI-powered sustainability software has expanded rapidly as organizations seek better ways to measure emissions, automate reporting and optimize operations. While features and pricing vary considerably, the following platforms, presented in alphabetical order, are among the leading options available to enterprise sustainability teams.</p> <h3><strong>CO2 AI</strong></h3> <p>CO2 AI focuses on enterprise carbon management and decarbonization planning. Its key components include product <a href="https://www.techtarget.com/sustainability/feature/How-businesses-can-track-carbon-emissions">carbon footprint calculations</a>, supply chain emissions management, reduction scenario modeling and AI-assisted emissions analysis.</p> <p>The platform is cloud-based and primarily targets large enterprises. <a target="_blank" href="https://co2ai.com/platform/platform-overview" rel="noopener">Pricing</a> is not readily available on the website, yet a free trial is offered.</p> <h3><strong>Diligent ESG</strong></h3> <p>Diligent ESG helps organizations collect ESG data, manage governance requirements and automate sustainability reporting across multiple frameworks. Its key capabilities include AI-assisted <a href="https://www.techtarget.com/sustainability/feature/Top-ESG-reporting-frameworks-explained-and-compared">ESG reporting</a>, regulatory compliance management, board and governance reporting and risk management dashboards.</p> <p>The platform is delivered as a cloud-based SaaS system, and <a target="_blank" href="https://www.diligent.com/pricing" rel="noopener">pricing</a> is available upon request.</p> <h3><strong>IBM Envizi</strong></h3> <p><a href="https://www.techtarget.com/searchbusinessanalytics/news/252511936/IBM-unveils-acquisition-of-Envizi-to-add-ESG-analytics-tools">IBM Envizi</a> is the vendor's ESG suite. It aims to help organizations collect sustainability data from multiple business systems while supporting ESG reporting and carbon accounting. Its core functions include automated data reporting, emissions tracking, ESG reporting and performance analytics.</p> <p>Envizi is available as a cloud service with enterprise <a target="_blank" href="https://www.ibm.com/products/envizi/pricing" rel="noopener">pricing</a>.</p> <h3><strong>Microsoft Cloud for Sustainability</strong></h3> <p>Microsoft Cloud for Sustainability integrates environmental data across Microsoft business applications and Azure services. Its key features include <a href="https://www.techtarget.com/sustainability/feature/The-top-carbon-accounting-software-for-2024">carbon accounting</a>, sustainability dashboards, emissions reporting and AI-powered data insights.</p> <p>The platform operates as a cloud service within the <a target="_blank" href="https://www.microsoft.com/en-us/sustainability#overview" rel="noopener">Microsoft ecosystem</a>.</p> <h3><strong>Persefoni</strong></h3> <p>Persefoni is an enterprise carbon accounting platform designed to measure, analyze and report greenhouse gas emissions. Its primary capabilities include carbon accounting, regulatory reporting, AI-assisted emissions analysis and management of <a href="https://www.techtarget.com/sustainability/feature/Scope-1-2-and-3-emissions-Differences-with-examples">Scope 1, 2 and 3 emissions</a>.</p> <p>Persefoni is available as a SaaS platform. <a target="_blank" href="https://www.persefoni.com/pricing" rel="noopener">Pricing</a> is available upon request, with a free trial available.</p> <h3><strong>Pulsora</strong></h3> <p>Pulsora combines sustainability data management with AI-powered ESG reporting and compliance capabilities. Its main features include ESG reporting automation, <a href="https://www.techtarget.com/sustainability/feature/How-to-build-a-sustainability-performance-management-stack">sustainability performance</a> dashboards, carbon accounting and regulatory compliance tracking.</p> <p>The <a target="_blank" href="https://www.pulsora.com/" rel="noopener">platform</a> is delivered as SaaS with enterprise licensing. Pricing is available upon request.</p> <h3><strong>Salesforce Net Zero Cloud</strong></h3> <p>Salesforce Net Zero Cloud manages emissions data alongside operational and customer information. Its primary functions include carbon accounting, supplier emissions tracking, ESG reporting and sustainability dashboards.</p> <p>The platform is offered as a SaaS application within Salesforce, and <a target="_blank" href="https://www.salesforce.com/net-zero/pricing/" rel="noopener">pricing</a> is available upon request.</p> <h3><strong>Watershed</strong></h3> <p>Watershed offers enterprise carbon accounting and <a href="https://www.techtarget.com/sustainability/feature/How-can-AI-help-sustainability">sustainability management with AI capabilities</a>. Its key functionality includes enterprise emissions management, supply chain carbon tracking, climate target management and reporting automation.</p> <p>Watershed is delivered as a cloud platform with custom <a target="_blank" href="https://watershed.com/" rel="noopener">enterprise pricing</a>.</p> </section> <section class="section main-article-chapter" data-menu-title="The environmental cost of AI"> <h2 class="section-title"><i class="icon" data-icon="1"></i><strong>The environmental cost of AI</strong></h2> <p>Although sustainability platforms continue to add GenAI capabilities, organizations must evaluate whether those features solve real business problems or simply automate existing workflows.</p> <p>Swanson recommended asking four questions before investing in AI initiatives:</p> <ul class="default-list"> <li>Is there a measurable physical outcome?</li> <li>Is the underlying data trustworthy?</li> <li>Will someone actually act on the AI's recommendations?</li> <li>And could a simpler technology deliver most of the same benefit?</li> </ul> <p>And despite AI's potential to improve sustainability performance, the <a href="https://www.techtarget.com/sustainability/feature/Generative-AIs-sustainability-problems-explained">technology itself carries environmental costs</a>.</p> <p>Training and operating large AI models requires enormous computing power. Expanding data center capacity increases electricity demand, while cooling systems consume substantial amounts of water. As organizations deploy AI more broadly, sustainability leaders must account for those effects alongside any emissions reductions AI enables elsewhere in the business.</p> <p>"The headline tradeoff is AI's own footprint," Swanson said. "Training and inference are energy and water-intensive."</p> <p>Many enterprise sustainability applications don't require the largest available models, she added.</p> <p>"Most enterprise sustainability use cases need a small model or classical optimization, not a frontier LLM," she said.</p> <p>Organizations often deploy resource-intensive models where smaller, more specialized models would produce comparable results using far less energy, Kazanov said.</p> <p>He also recommends considering when AI workloads run, suggesting organizations <a href="https://www.techtarget.com/searchnetworking/tip/How-to-modernize-networks-with-sustainability-in-mind">schedule compute-intensive processing</a> during periods when renewable energy contributes more heavily to the electrical grid.</p> <p>Measuring AI's environmental impact requires looking beyond traditional data center efficiency metrics such as power usage effectiveness, Kazanov said. Organizations should also track marginal carbon intensity, water usage, embodied carbon in AI hardware and the full lifecycle of specialized computing equipment.</p> <p>Organizational accountability presents another challenge, said Arif Gasilov, partner for natural resources and the built environment at Gasilov Group.</p> <p>"IT enables access, departments pay for licensing, facilities pays the electric bill [and] sustainability reports on emissions," he said.</p> <p>As a result, no single group fully owns AI's environmental costs.</p> <p>Organizations should measure the overall effect of their AI infrastructure on the broader energy system rather than focusing solely on internal sustainability metrics, Gasilov said. This can involve calculating AI's net environmental impact by comparing the emissions reductions it enables against the energy and resources required to build and operate the technology, Swanson said.</p> <p>Her final metric may be the simplest -- and the most revealing.</p> <p>"Measure whether anyone acts on the model's output," she said. "A model nobody uses is pure cost, carbon included."</p> <p><em>Christine Campbell is a freelance writer specializing in business and B2B technology.</em></p> </section> AI tools can enhance sustainability with optimized energy use, improved predictive maintenance and streamlined supply chains while managing their own environmental impact. https://cdn.ttgtmedia.com/rms/onlineimages/esg_a488313427.jpg https://www.techtarget.com/it-strategy/feature/Most-promising-AI-tools-for-sustainability-professionals Wed, 29 Jul 2026 12:00:00 GMT <p>Nimble on Wednesday launched Web Search Agents, new agentic AI applications that autonomously learn customer use cases to scour the internet for relevant information that can be converted into curated data tables.</p> <p>While analytics and AI tools are largely informed by an enterprise's operational data, information from the web is also valuable when building applications that inform decisions and business processes. Operational data reveals what is happening within an organization, while web data shows what is taking place externally.</p> <p>Together, the two provide <a href="https://www.techtarget.com/searchdatamanagement/opinion/Context-is-the-make-or-break-layer-for-AI-in-production">richer context for agents</a> and other decision-informing tools than one type of data alone.</p> <p>"Web data is the world's largest real-time signal network," Michael Ni, an analyst at Constellation Research, told TechTarget. "Nimble allows enterprises to combine internal and external data, enabling AI to move from internally informed analysis to context-aware decisions."</p> <p>Use cases include real time competitive intelligence, such as dynamic pricing and product availability, examining market and sentiment signals, insight into competitors, and <a href="https://www.techtarget.com/searchenterpriseai/tip/Global-AI-legislation-and-regulation-tracker">ensuring regulatory compliance</a>.</p> <p>Nimble's Web Search Agents are designed to automate much of the research work that previously had to be performed by humans, building and maintaining complex <a href="https://www.techtarget.com/whatis/feature/How-to-scrape-data-from-a-website">web-scraping tools</a>. In addition, Nimble's agents are designed to self-learn a user's domain to perform web data research with the accuracy of an expert, but with greater efficiency.&nbsp;&nbsp;</p> <p>As a result, Web Search Agents are a potentially significant addition for Nimble customers, according to Donald Farmer, founder and principal of TreeHive Strategy.</p> <p>"There has always been a bottleneck accessing web data because it meant building and maintaining brittle scraping scripts and managing proxies," he told TechTarget. "Agents should autonomously … remove that bottleneck."</p> <p>Based in New York City, Nimble is a web data search specialist that <a href="https://www.techtarget.com/searchdatamanagement/news/366639416/Web-data-curator-Nimble-secures-47M-to-fuel-growth">raised $47 million</a> in venture capital funding in February to fuel its growth. In addition to Nimble, vendors that specialize in enabling users to pull data from the web include Apify, Bright Data, Grepsr, MixRank and&nbsp;<a href="https://www.computerweekly.com/blog/Data-Matters/Giving-ourselves-the-chance-to-lead-the-AI-race-and-stay-ethical-too">Oxylabs</a>.</p> <section class="section main-article-chapter" data-menu-title="Aiding AI"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Aiding AI</h2> <p>Agents and other AI tools require reams of contextually relevant information to deliver accurate outputs at a high enough rate that they can be <a href="https://www.techtarget.com/searchcio/feature/Startup-founder-says-trust-is-biggest-barrier-to-AI-agents">trusted in production</a>.</p> <p>Despite continually <a href="https://kpmg.com/us/en/media/news/q1-ai-pulse2026.html">rising investments</a> in AI development, <a href="https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html?id=us:2ps:3gl:aisgm26:awa:CONS:em:K0218784:012626:kwd-430833501819:195648817121:794247818306::&amp;gclsrc=aw.ds&amp;gad_source=1&amp;gad_campaignid=23269751971&amp;gclid=CjwKCAjw6MPRBhBTEiwAd-7Mrwla_c_gzzSu2TNPVBWdBS4INV4yNdYgtukO2OU2-zWBAuwZE-s3mBoCwegQAvD_BwE">more projects fail</a> than succeed. Frequently, difficulties discovering and connecting agents with the context that enables them to perform properly &nbsp;prevent AI initiatives from moving beyond the pilot stage.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> There has always been a bottleneck accessing web data because it meant building and maintaining brittle scraping scripts and managing proxies. Agents should autonomously … remove that bottleneck. </figure> <figcaption> <strong>Donald Farmer</strong>Founder and principal, TreeHive Strategy </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>With context critical to developing trustworthy AI tools, many vendors have introduced products in 2026 designed to help customers discover and operationalize relevant data. In June alone, <a href="https://www.techtarget.com/searchdatamanagement/news/366644853/AWS-latest-to-introduce-context-layer-for-agentic-AI">AWS</a>, <a href="https://www.techtarget.com/searchdatamanagement/news/366644813/Databricks-intros-new-Genie-data-management-tools-to-aid-AI">Databricks</a>, <a href="https://www.techtarget.com/searchdatamanagement/news/366643955/Microsoft-boosts-Fabric-to-make-it-a-foundation-for-AI">Microsoft</a> and <a href="https://www.techtarget.com/searchdatamanagement/news/366643795/Snowflake-barrage-adds-more-AI-development-analysis-tools">Snowflake</a> unveiled new features aimed at connecting agents with context.</p> <p>With Web Data Agents, Nimble is similarly providing capabilities aimed at helping customers feed agents with contextually relevant data, in a move motivated by user feedback, according to Uri Knorovich, the vendor's co-founder and CEO.</p> <p>"The feedback we kept hearing was simple -- [customers'] agents could search the web, but they weren't getting the specific data they needed. … Agents need web search that adapts to their specific task, context and objectives to optimize for accuracy and cost, not one-size-fits-all results," he said.</p> <p>Web Search Agents are built on customized search algorithms that enable them to learn autonomously, Knorovich continued.</p> <p>"These advances allow search models to continuously learn a domain, optimize retrieval for a specific use case and deliver … more accurate web context than generic search engines," he said.</p> <p>Beyond accurately finding and retrieving relevant web data, Nimble's Web Search Agents are designed to <a href="https://www.techtarget.com/searchenterpriseai/tip/Reduce-project-expenses-with-AI-cost-optimization-strategies">reduce the cost</a> of searching the internet for task-specific context.</p> <p>Using generic search engines that are task-agnostic can become expensive, according to Farmer, who noted that such tools often return massive, unstructured amounts of text that, when processed by an agent, consumes a high number of <a href="https://www.techtarget.com/whatis/definition/context-window">context window</a> tokens.</p> <p>In benchmark testing conducted by Nimble but not independently verified, Web Search Agents increased answer quality by over 20% and approximately halved the number of tokens spent on each query.</p> <p>"A smart agent retrieving data should therefore be a great advantage," Farmer said.</p> <p>However, while valuable for Nimble customers, Web Search Agents are not the only such web data discovery and retrieval tools on the market, he continued.</p> <p>For example, Grepsr and MixRank both offer AI-powered web scraping capabilities that can discover and deliver task-specific data. However, Farmer pointed out that there is slight differentiation, since Nimble's Web Search Agents deploy <a href="https://www.techtarget.com/whatis/definition/headless-browser">headless browsers</a> that pull the freshest version of a web page instead of a data <a href="https://www.techtarget.com/searchstorage/definition/cache">cache</a>.</p> <p>"If you need that, it is a compelling feature," Farmer said. "I don't think there's a lot here that is innovative compared to the rest of the market, but the work appears to be solid."</p> <p>Ni similarly noted that Nimble's web data collection capabilities are not unique. Instead, the vendor has the potential to distinguish itself by combining live access to web data, agent-powered navigation and repeatable output into a managed external sensing layer, he continued.</p> <p>"The proof point will be whether that delivers greater completeness and lower maintenance costs than traditional scraping platforms, vertical data platforms, or AI-native search APIs," Ni said.</p> </section> <section class="section main-article-chapter" data-menu-title="Next steps"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Next steps</h2> <p>While operational data remains the focal point for many enterprises, Nimble's leadership believes that web data is becoming the primary source of <a href="https://www.techtarget.com/searchenterpriseai/tip/Exploring-the-context-layer-for-AI-systems">context for agents</a>, according to Knorovich.</p> <p>With that in mind, Nimble plans to make it easier to connect an enterprise's agents with web search by adding and improving self-learning and agent memory capabilities so Web Search Agents better understandtasks and get smarter with each use.</p> <p>"The bigger vision is a web that agents can navigate as fluently as people do, and we intend to keep building toward that," Knorovich said.</p> <p>Ni noted that Nimble is positioning itself as an enterprise-grade layer for providing external context to agents. Going forward, its opportunity to continue serving existing users and attract new ones is by turning that layer into <a target="_blank" href="https://www.linkedin.com/pulse/rise-ai-control-planes-governing-models-agents-scale-mahmoud-abufadda-d1f5f/" rel="noopener">a control plane</a> for external context.</p> <p>"Parallel to the growth of internal context layers, enterprise leaders will want to know where any specific fact came from, how fresh and reliable it is, whether sources conflict, and whether an agent is permitted to act on it given a specific context," Ni said. "Productizing context controls would expand Nimble’s relevance from developers to CDAOs and CAIOs."</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> The vendor's Web Search Agents autonomously learn domain-specific use cases to help enterprises collect contextually relevant information for AI and analytics. https://cdn.ttgtmedia.com/rms/onlineimages/telecommunications_g1189468316.jpg https://www.techtarget.com/data-technologies/news/366646240/Nimble-launches-agents-to-cull-relevant-web-data-for-AI Wed, 29 Jul 2026 09:00:00 GMT <p>Lack of context is the enemy of accuracy.</p> <p>After <a href="https://www.techtarget.com/searchdatamanagement/feature/Data-quality-fast-failures-and-quick-wins-key-to-AI-success">experimenting with agentic AI</a>, many enterprises are attempting to move projects past the pilot stage and into production to finally realize the productivity gains and <a href="https://www.techtarget.com/searcherp/feature/AI-feature-spend-is-the-new-software-cost-control-problem">cost savings</a> of deploying agentic AI networks.</p> <p>However, to reap the potential rewards of agentic AI and avoid serious consequences, organizations need to make sure they're developing agents that can be trusted to perform in production.</p> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/farmer_donald.jpg" alt="Donald Farmer, founder and principal of TreeHive Strategy">Donald Farmer </div> <p>Enterprises need to provide AI tools with the context -- the data and <a href="https://www.techtarget.com/whatis/definition/business-logic">business logic</a> -- that enables them to accurately carry out their intended work. If they don't, agents could wind up costing businesses time, &nbsp;money and, in some cases, reputational harm or legal peril.</p> <p>"With good context, an agent can map vague natural language requests to exact corporate data," Donald Farmer, founder and principal of TreeHive Strategy, told TechTarget. "Without it, the agent can only act generically and may hallucinate what it does not know."</p> <section class="section main-article-chapter" data-menu-title="Investments in AI continue to rise"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Investments in AI continue to rise</h2> <p>In January, research and advisory firm <a href="https://www.gartner.com/en/newsroom/press-releases/2026-1-15-gartner-says-worldwide-ai-spending-will-total-2-point-5-trillion-dollars-in-2026" target="_blank" rel="noopener">Gartner predicted</a> that worldwide spending on AI will total $2.5 trillion in 2026, up 44% from $1.8 trillion in 2025, and increase to $3.3 trillion in 2027.</p> <p>Simultaneously, some enterprises are putting at least some agents into production as they try to benefit from AI's potential.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> With good context, an agent can map vague natural language requests to exact corporate data. Without it, the agent can only act generically and may hallucinate what it does not know. </figure> <figcaption> <strong>Donald Farmer</strong>Founder and principal, TreeHive Strategy </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>ISG Software Research's year-end 2024 survey found that at the time, only 17% of AI initiatives reached production, according to ISG analyst David Meninger. One year later, the firm's year-end 2025 survey showed about one-third of AI projects making it into production.</p> <p>Similarly, <a href="https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html?id=us:2ps:3gl:aisgm26:awa:CONS:em:K0218784:012626:kwd-430833501819:195648817121:794247818306::&amp;gclsrc=aw.ds&amp;gad_source=1&amp;gad_campaignid=23269751971&amp;gclid=CjwKCAjw6MPRBhBTEiwAd-7Mrwla_c_gzzSu2TNPVBWdBS4INV4yNdYgtukO2OU2-zWBAuwZE-s3mBoCwegQAvD_BwE" target="_blank" rel="noopener">Deloitte's State of AI in the Enterprise report</a>, published in January, found that one-quarter of the 3,200 business and IT leaders surveyed reported having put 40% of their AI projects into production. In addition, Deloitte predicted that about half of the organizations it surveyed would reach that 40% threshold by midyear.</p> <p>"We are making significant progress with agents -- we have more enterprises putting agents into production -- but that still leaves a lot that aren't in production," Menninger said. "If we are doubling the number of agents going into production, we're paying more attention to the issues surrounding production quality agents."</p> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/menninger_david.jpg" alt="ISG Software Research analyst David Menninger">David Menninger </div> <p>In other words, many organizations are paying more attention to context to realize agentic AI's potential. However, it's imperative that they get the context aspect of development correct, Menninger continued.</p> <p>If agents are asked to execute straightforward tasks such as drafting a document, accuracy isn't imperative. Mistakes can be caught, and the consequences aren't catastrophic. But if <a href="https://www.techtarget.com/searchapparchitecture/tip/Multi-agent-architectures-How-coordinated-AI-systems-scale">multi-agent systems</a> are managing entire supply chains or making <a href="https://www.techtarget.com/healthtechanalytics/feature/LLMs-struggle-with-clinical-reasoning-study-finds">recommendations to medical workers</a>, government employees or other people doing critical work, the consequences of agents acting without proper context can be severe.</p> <p>"Now that we're starting to do more agentic activities … we need to get it right, and the risks are higher," Menninger said.</p> </section> <section class="section main-article-chapter" data-menu-title="Real-world consequences"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Real-world consequences</h2> <p>Enterprises are spending vast amounts of money and devoting significant time and effort to building agents because the potential benefits of deploying networks of AI agents are massive.</p> <p>For example, <a href="https://arxiv.org/pdf/2503.18238" target="_blank" rel="noopener">authors from Johns Hopkins and MIT</a> reported that teams comprised of humans and agents were 60% more productive than teams made up of humans alone. Meanwhile, <a href="https://www.randgroup.com/insights/services/ai-machine-learning/how-much-does-ai-save-a-company/">a separate study</a> conducted by Rand Group found that agents can help enterprises reduce spending by 40%.</p> <p>But there are repercussions when AI tools lack appropriate context.</p> <p>"Context is given to the agent to ensure accuracy," Cindi Howson, chief data and AI strategist at ThoughtSpot, told TechTarget. "In the absence of context, you get hallucinations. The agent takes its best guess, but it is just a guess."</p> <p>Real-world <a href="https://www.evidentlyai.com/blog/ai-hallucinations-examples" target="_blank" rel="noopener">examples of AI hallucinations</a> causing harm include the following:</p> <ul type="disc" class="default-list"> <li>Air Canada suffered reputational harm and was forced to refund money when an AI chatbot didn't know the company's policy related to bereavement travel and misled a passenger.</li> <li>Google's parent company, Alphabet, lost $100 billion in market value after incorrect AI-generated content was included in a promotional video.</li> <li>OpenAI's Whisper, used by tens of thousands of medical professionals to transcribe patient visits, made up entire sentences, invented medication names and attributed racially charged statements to patients with aphasia, a disorder that impairs the ability to speak.</li> <li>A California court found that agents from Perplexity <a href="https://www.reuters.com/legal/litigation/amazon-wins-order-blocking-access-perplexitys-ai-shopping-agent-2026-03-10/" target="_blank" rel="noopener">may be violating state and federal laws</a> by accessing Amazon accounts without prior authorization.</li> </ul> <p>Beyond public consequences, there are private ones when agents don't have the context they need to deliver accurate outputs, Howson noted, pointing out that organizations are charged for LLM use by the number of <a href="https://www.theserverside.com/tutorial/An-introduction-to-LLM-tokenization">tokens they consume</a>.</p> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/howson_cindi.jpeg" alt="Cindi Howson, chief data and AI strategist at ThoughtSpot">Cindi Howson </div> <p>"Happening in parallel are the token costs," she said. "The more context you give an LLM, you're going to get a better answer on the first attempt, whereas if the LLM gets it wrong, you're going to run up your token costs and it's not going to be efficient."</p> <p>When an agent cannot be trusted in production, the time and money spent building it will be wasted. Not including all the tools it takes to create an AI infrastructure, the cost to develop a single agent can range from around $1,000 for simple applications to over $100,000, according to IT consulting firm Triple Minds.</p> <p>Differences come down to the scope, autonomy level and <a href="https://www.techtarget.com/searcherp/podcast/Plan-a-multi-agent-orchestration-framework-for-scalable-AI">depth of integration</a> with enterprise systems, including other agents. A basic agent that responds to frequently asked questions might cost around $15,000 to build, while a custom-trained, fully autonomous agent capable of working with other agents could cost over $100,000.</p> <p>"Enterprises want agents that are reliable, with responses that are grounded, properly attributed and consistent in how long they take to run," Michael Bendersky, director of research at Databricks, told TechTarget. "Speed and cost matter as much as accuracy. An agent that [takes] unnecessary steps before reaching the right answer is expensive and unpredictable at scale."</p> </section> <section class="section main-article-chapter" data-menu-title="Success stories"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Success stories</h2> <p>While failure to connect agents with appropriate context can have consequences, the potential benefits of agentic AI outweigh the risks. Improved efficiency and cost savings are real when agents can instantaneously access the relevant data and business logic they need to be accurate.</p> <p>Over the past few months, as data management and analytics providers have hosted their annual user conferences, they've put on a parade of companies that are among those having early success building, deploying and benefiting from agents.</p> <p>In late March, during one of the spring's first conferences, <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Domo-drives-customer-excitement-with-latest-AI-capabilities">analytics vendor Domo</a> introduced Nichole Gunn, CEO of channel marketing specialist Extu. Aided by a specialist from Domo, Gunn used capabilities launched by the vendor during its conference to build an agent in less than 30 minutes that takes on time-consuming operational tasks related to Extu's customers.</p> <p>Gunn, who wanted an agent that eliminated multi-step onboarding tasks that must be repeated for each Extu customer, estimated that the agent <a href="https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/agents-for-growth-turning-ai-promise-into-impact" target="_blank" rel="noopener">would save</a>&nbsp;Extu hundreds of thousands of dollars in operational costs in its first few months in production and millions within six months.</p> <p>The agent not only handles onboarding but provides updates on customers and suggests campaigns to keep customers and partners engaged, according to Gunn.</p> <p>"It went from this one thing that I wanted to a multi-faceted way of saving our company a lot of money," she told TechTarget in March. "We will be using the AI agents immediately when I get home."</p> <p>Throughout April and May, vendors including Alteryx, Informatica, Qlik and Tableau held events to showcase success stories. And just last month, AWS, Databricks, Microsoft and Snowflake closed out the spring conference season with events where they highlighted customers successfully building contextually aware agents and realizing some of agentic AI's potential.</p> <p>One of the customers Snowflake highlighted was <a href="https://www.computerweekly.com/news/366642986/Accenture-joins-IBM-in-battle-for-323m-Post-Office-Horizon-deal">Accenture</a>, a global IT services and consulting firm that is using <a href="https://www.techtarget.com/searchdatamanagement/news/366643795/Snowflake-barrage-adds-more-AI-development-analysis-tools">Snowflake's platform</a> to help its own customers manage their data and build insight-generating analytics and AI tools.</p> <p>For example, a European utility company is using combined capabilities from Snowflake and Accenture to reduce queries from weeks to seconds, according to Accenture chief strategy and services officer Manish Sharma, who spoke with <a href="https://www.techtarget.com/searchbusinessanalytics/news/366571855/Snowflake-CEO-Slootman-steps-down-Ramaswamy-takes-over">Snowflake CEO Sridhar Ramaswamy</a> during the Snowflake Summit keynote address on June 1. Similarly, a U.S. manufacturer is using tools from Snowflake and Accenture to create a unified data layer for analytics and AI after its data was previously fragmented across myriad systems.</p> <p>"There's a simplicity and beauty to all of this," Ramaswamy said. "We succeed when we help our customers make more money or spend less money."</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> When connected with relevant context, agents can increase organizational efficiency and reduce spending. When left to guess, they can cause significant harm. https://cdn.ttgtmedia.com/rms/onlineimages/machine_learning_g1209661950.jpg https://www.techtarget.com/data-technologies/news/366646162/Potential-consequences-are-severe-when-AI-agents-lack-context Wed, 29 Jul 2026 08:00:00 GMT <p>The White House Executive Order 14409, "Promoting Advanced Artificial Intelligence Innovation and Security," issued in June 2026, creates a voluntary engagement model between frontier AI developers and the federal government.</p> <p>For CIOs, CTOs, CISOs and procurement leaders, it contains signals useful for AI vendor due diligence. If a vendor develops, deploys or resells access to frontier AI models, <a href="https://www.techtarget.com/searchenterpriseai/news/366644013/Trump-AI-order-targets-frontier-model-prerelease-review">executive order (EO) compliance</a> becomes a procurement issue even when the buyer is not directly regulated, because buyers will want to know whether the vendor understands the risk class of its models.</p> <section class="section main-article-chapter" data-menu-title="How the executive order affects vendors"> <h2 class="section-title"><i class="icon" data-icon="1"></i>How the executive order affects vendors</h2> <p>The EO starts from a cybersecurity premise: that advanced AI can help defend systems, find vulnerabilities and support critical infrastructure, but that the <a href="https://www.techtarget.com/searchcio/feature/ais-cybersecurity-paradox-how-CIOs-can-keep-up-with-change">same capabilities can create national security risks</a>.</p> <p>The key vendor management provision is Section 3, "Secure Frontier Model Deployment." It directs the Treasury, the Secretary of War through the National Security Agency, and Homeland Security through CISA, with NIST and other federal offices, to develop a classified benchmarking process for advanced cyber capabilities in AI models. The process determines when a model becomes a covered frontier model -- a risk threshold tied to advanced cyber capability. Because the threshold is classified, enterprises should not expect vendors to disclose the benchmark.</p> <p>The framework is voluntary. Developers may ask the federal government to assess whether models under development meet the designation and may provide federal access to covered frontier models for up to 30 days before release to other trusted partners, subject to confidentiality, cybersecurity, insider risk, intellectual property, use and nondisclosure protections. This creates an expectation of transparency without a formal approval stamp.</p> <p>Vendors should be able to state whether they participate, whether a model has been submitted, what government access and confidentiality terms apply and how they manage advanced cyber capability risk. The 30-day window also signals maturity: The vendor can control release candidates, support secure evaluation, protect model weights, respond to findings and manage release discipline.</p> <div class="youtube-iframe-container"> <iframe id="ytplayer-0" src="https://www.youtube.com/embed/Xvj-dz30GiY?autoplay=0&amp;modestbranding=1&amp;rel=0&amp;widget_referrer=null&amp;enablejsapi=1&amp;origin=https://www.techtarget.com" type="text/html" height="360" width="640" frameborder="0"></iframe> </div> <p>In parallel, <a href="https://www.techtarget.com/searchsecurity/news/366645907/Industry-reacts-to-Gold-Eagle-vulnerability-management-plan">Gold Eagle</a>, the White House's AI cybersecurity vulnerability coordination <a target="_blank" href="https://www.whitehouse.gov/releases/2026/07/white-house-launches-gold-eagle-initiative-for-unprecedented-cybersecurity-vulnerability-coordination/" rel="noopener">initiative</a> established in the June EO, points the same way. Vendors are now evaluated on coordinated vulnerability management alongside model quality and feature velocity -- the rate at which new features are brought to users.</p> <p>Some of the evaluation points include the following:</p> <ul class="default-list"> <li>Vulnerability intake.</li> <li>Customer notification.</li> <li>Patch coordination.</li> <li>Protection of sensitive vulnerability information.</li> <li>Collaboration with agencies, open source maintainers and critical infrastructure operators.</li> </ul> </section> <section class="section main-article-chapter" data-menu-title="Considerations for AI vendor evaluation"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Considerations for AI vendor evaluation</h2> <p>CIOs should not treat the EO as only a federal procurement document. The order mainly directs federal agencies. Participation is voluntary, but standards often shape vendor expectations before they become legal obligations, as <a href="https://www.techtarget.com/searchsecurity/definition/Soc-2-Service-Organization-Control-2">SOC 2</a> did, and frontier AI testing is starting down the same path.</p> <p>For procurement teams, the EO's "advanced cyber capabilities" should translate into practical due diligence. Some questions to ask include the following:</p> <ul class="default-list"> <li>Can the <a href="https://www.techtarget.com/searchenterpriseai/news/366642478/Claude-Mythos-Preview-and-the-new-rules-of-cybersecurity">model autonomously discover vulnerabilities</a>?</li> <li>Can it chain findings across systems?</li> <li>Can it generate exploit-relevant code or operational guidance?</li> <li>Can it interact with repositories, terminals, ticketing systems, tools or production environments?</li> <li>Can it make changes without meaningful human review?</li> <li>Can outputs help defenders while also increasing attacker capability?</li> </ul> <p>These answers determine whether a system is made available in a sandbox, development environment, SOC workflow, production network or a more restricted setting.</p> <p>CIOs should require structured attestations on frontier-risk assessment, voluntary-framework engagement, cybersecurity capability evaluation and procedures to withhold, roll back or constrain a model after evaluation.</p> <p>Red-teaming should become a default procurement requirement. NIST's AI Risk Management Framework helps identify, measure, manage and govern AI risk. Its generative AI profile, <a target="_blank" href="https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf" rel="noopener">NIST AI 600-1</a>, helps select actions aligned to organizational goals.</p> <p>Scope should cover the following:</p> <ul class="default-list"> <li>Prompt injection.</li> <li>Insecure output handling.</li> <li>Training data poisoning.</li> <li>Model denial-of-service.</li> <li><a href="https://www.techtarget.com/searcherp/feature/5-supply-chain-cybersecurity-risks-and-best-practices">Supply chain vulnerabilities</a>.</li> <li>Sensitive information disclosure.</li> <li>Insecure plugin design.</li> <li>Excessive agency.</li> <li>Overreliance.</li> <li>Unsafe code generation.</li> <li>Model theft.</li> </ul> <p>Testing should reflect the actual workflow, including access to email, code repositories, customer records, financial and HR systems, identity stores, APIs and operational technology.</p> <p>For high-impact uses, such as employment, lending, healthcare, insurance, education or government benefits, vendors should provide documentation supporting compliance with state laws.</p> <p>For example, Colorado's revised <a target="_blank" href="https://leg.colorado.gov/bills/sb26-189" rel="noopener">SB26-189</a>, effective January 2027, requires developers of covered automated decision-making technology to give deployers documentation on intended uses, training-data categories, known limitations, appropriate use and human review, and material updates.</p> <p>California's <a target="_blank" href="https://www.gov.ca.gov/2025/09/29/governor-newsom-signs-sb-53-advancing-californias-world-leading-artificial-intelligence-industry/" rel="noopener">SB 53</a> requires large frontier AI developers to publish a framework that incorporates national, international and industry consensus standards, with mechanisms to report critical safety incidents and protect whistleblowers.</p> <p>Global vendors face overlapping obligations. Under the EU AI Act, general-purpose AI model provider obligations apply since August 2025, European Commission enforcement begins in August 2026, and models on the market must comply by August 2027. So, a single certification may not suffice.</p> <p><a href="https://www.techtarget.com/searchenterpriseai/feature/AI-regulation-What-businesses-need-to-know">Compliance obligations vary</a> based on who built the model, where it is deployed, what it decides, whether it is general-purpose and whether it sits in a high-impact workflow or has systemic or frontier capabilities.</p> <p>Intellectual property and confidentiality need separate attention. If a vendor seeks access to customer source code, internal tickets, vulnerability reports, proprietary documents or sensitive prompts, the contract should define the following:</p> <ul class="default-list"> <li>Ownership.</li> <li>Retention.</li> <li>Training restrictions.</li> <li>Logging.</li> <li>Deletion.</li> <li>Subcontractor access.</li> <li>Audit rights.</li> </ul> <p>Vendors will also protect model weights, prompts, evaluations and system instructions as proprietary, so procurement teams need secure review processes that respect confidentiality without accepting a black box.</p> </section> <section class="section main-article-chapter" data-menu-title="Practical vendor due diligence checklist"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Practical vendor due diligence checklist</h2> <p>Diligence is required through all steps of the procurement process, from pre-RFP assessment through the RFP, contract negotiation and ongoing AI vendor management.</p> <p>The following checklist translates the EO and <a href="https://www.techtarget.com/searchsecurity/tip/What-CISOs-need-to-know-about-AI-governance-frameworks">related governance signals</a> into a due diligence process that can anchor AI procurement strategies.</p> <h3>Pre-RFP assessment</h3> <p>Classify the vendor and AI capability before writing the RFP. A vendor embedding a third-party chatbot in a narrow workflow does not create the same risk as one that develops frontier models, deploys autonomous agents or sells AI-enabled vulnerability discovery.</p> <p>Key steps at this stage of assessment:</p> <ul class="default-list"> <li>Determine whether the vendor develops, fine-tunes, hosts, resells or deploys frontier AI models, and whether it is the model developer, application-layer provider, systems integrator, hosting provider or reseller.</li> <li>Assess whether the model could plausibly fall within the EO's covered-frontier-model concern and whether the vendor participates in, plans to join or is monitoring the voluntary review framework, keeping participation, nonparticipation and formal approval distinct.</li> <li>Review the vendor's history of <a href="https://www.techtarget.com/searchsecurity/feature/What-to-know-about-red-team-testing-and-the-law">adversarial testing against the red-team scope</a> described above.</li> <li>Review state and international compliance posture against the laws noted above, plus other transparency and sector-specific rules.</li> <li>Map the intended enterprise use. Summarization, autonomous remediation, invoice approval, candidate screening and production code generation carry different risk profiles.</li> </ul> <h3>RFP requirements</h3> <p>Compliance documentation should go beyond standard security questionnaires. The RFP should require the following structured model and system documentation:</p> <ul class="default-list"> <li>Model family.</li> <li>Deployment model.</li> <li>Hosting location.</li> <li>Versioning.</li> <li>Fine-tuning.</li> <li>Tool integrations.</li> <li>Retrieval sources.</li> <li>Data flows.</li> <li><a href="https://www.techtarget.com/searchenterpriseai/feature/Humans-and-AI-The-role-of-people-in-the-new-AI-world">Human review points</a>.</li> <li>Fallback modes.</li> </ul> <p>It should also require the following:</p> <ul class="default-list"> <li>Capability documentation, including supported and prohibited tasks, known limitations, benchmark summaries, safety evaluations and cybersecurity-relevant capabilities.</li> <li>Evidence of vulnerability assessments, including internal and third-party testing, remediation history, disclosure policies, patch cadence, dependency management, secure development and model-weight controls.</li> <li>Training data transparency sufficient for legal and operational review.</li> <li><a href="https://www.techtarget.com/searchcustomerexperience/answer/How-do-companies-protect-customer-data">Customer data protection</a> terms prohibiting use of inputs, outputs, files, logs, prompts, embeddings or derived data for training or product improvement without written consent, plus clear telemetry rules for security and abuse monitoring.</li> <li>Update and patch protocols covering testing, notification, version documentation, version pinning and emergency rollback.</li> <li>Audit evidence, including SOC 2 Type II, ISO 27001, ISO/IEC 42001, penetration-test and AI safety assessment summaries, vulnerability metrics, subcontractor attestations and alignment to NIST AI RMF or comparable frameworks.</li> <li>Explainability appropriate to the use case.</li> </ul> <h3>Contract negotiation</h3> <p>Marketing claims and RFP responses should become binding obligations in exhibits, schedules or security addenda.</p> <p>Key contractual terms include the following:</p> <ul class="default-list"> <li>Prohibition on customer data use for training, fine-tuning, evaluation, benchmarking or product improvement without prior written consent, covering prompts, outputs, uploaded files, embeddings, logs, metadata, telemetry and derived data, including subcontractor use.</li> <li>Incident notice and government reporting obligations when an incident affects the customer's environment, data, operations or compliance.</li> <li><a href="https://www.techtarget.com/searchsecurity/news/366644312/Its-time-to-update-incident-response-for-the-AI-era">AI-specific incident response</a> commitments for prompt injection, unauthorized model actions, output-based data leakage, model-weight or system-prompt compromise, unsafe autonomous behavior and harmful erroneous outputs.</li> <li>Audit rights and third-party assessment access for high-risk deployments, plus rollback and version-control procedures.</li> <li>Liability allocation for product defects, control failures, unauthorized data use, IP infringement, security failures and missed compliance commitments.</li> <li>Change-control obligations for material changes to model capabilities, training practices, hosting architecture, sub-processors, safety filters, logging, retention, cybersecurity controls or government-review status.</li> <li>IP and confidentiality protections in both directions, not just for the vendor.</li> </ul> <h3>Ongoing vendor management</h3> <p><a href="https://www.techtarget.com/searchenterpriseai/news/366644404/MIT-study-warns-of-major-AI-risk-Is-governance-keeping-up">AI risks evolve</a> as models, integrations, security threats and laws change. A model acceptable in the past may look different in the future after a capability jump, a new tool integration or a legal change. CIOs should set a quarterly review cadence for material AI vendors, covering model updates, new capabilities, incidents, vulnerability findings, red-team results, regulatory and sub-processor changes, customer data use and SLA performance.</p> <p>Ongoing management should also include the following:</p> <ul class="default-list"> <li>Monitor participation in voluntary federal processes, Gold Eagle or related coordination efforts, and whether government-facing findings triggered product changes.</li> <li>Review new model versions, agentic tools, code-execution features, retrieval connectors and administrative permissions before deployment.</li> <li>Maintain <a href="https://www.techtarget.com/searchdisasterrecovery/tip/Why-a-risk-management-framework-is-critical-for-AI-initiatives">updated risk assessments</a> across cybersecurity, data protection, legal compliance, operational resilience, model behavior, IP exposure and business continuity, distinguishing vendor-level from use case-level risk.</li> <li>Run periodic red-team or tabletop exercises for important deployments and reassess state and international obligations at least twice a year.</li> </ul> <p>For CIOs, the main point is vendor classification by capability, deployment role and risk pathway. A frontier-model provider, a SaaS vendor with an embedded model, a workflow automation vendor and an AI cybersecurity platform are different kinds of suppliers.</p> <p>The EO is voluntary, but vendors that document model capabilities, testing practices, data restrictions, update controls and incident obligations will be easier to trust. Vendors that cannot are, in effect, shifting hidden risk to the buyer.</p> <p><em>Kashyap Kompella, founder of RPA2AI Research, is an AI industry analyst and advisor to leading companies across the U.S., Europe and the Asia-Pacific region. Kashyap is the co-author of three books,&nbsp;Practical Artificial Intelligence,&nbsp;Artificial Intelligence for Lawyers and AI Governance and Regulation.</em></p> </section> June's AI Executive Order promotes voluntary engagement between AI developers and the federal government, emphasizing vendor management and advanced cybersecurity standards. https://cdn.ttgtmedia.com/rms/onlineimages/collab_a275903017.jpg https://www.techtarget.com/it-strategy/tip/How-the-AI-Executive-Order-shifts-vendor-management-strategies Tue, 28 Jul 2026 14:00:00 GMT <p>The announcements have cascaded like an avalanche.</p> <p>One after another throughout the first six months of 2026, data management vendors unveiled and launched new capabilities designed to connect AI agents with the context needed to perform in production.</p> <p>Just days into January, Databricks and MongoDB each introduced new capabilities aimed at discovering and delivering agents the relevant data and business logic they require. Since then, most product development releases have similarly focused on enabling AI tools to call on contextually appropriate information the instant they need it.</p> <p>The reason is simple: without appropriate context, AI agents will fail.</p> <p>"Without it, they're guessing," Michael Bendersky, director of research at Databricks, told TechTarget. "Guesswork doesn't work for enterprise businesses."</p> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/bendersky_michael.jpg" alt="Michael Bendersky, director of research at Databricks">Michael Bendersky </div> <p>Agents always attempt to answer questions and act when asked, whether they have the information to do so accurately or not. They make <a href="https://www.techtarget.com/whatis/definition/What-is-AI-inference">inferences</a> based on the context they can access, and if that context isn't relevant, their inference won't be relevant either.</p> <p>Connecting agents with relevant context, however, is complex.</p> <p>Data and <a href="https://www.techtarget.com/whatis/definition/business-logic">business logic</a> need to be prepared for AI, which can be a lengthy, labor-intensive process for massive organizations with data spread across disparate systems. And relevant context needs to be instantly discoverable and retrievable. Otherwise, <a target="_blank" href="https://business.uq.edu.au/momentum/why-80-per-cent-ai-projects-fail" rel="noopener">agent initiatives will break down</a>, either never making it past the pilot stage or being untrustworthy -- and unusable -- in production.</p> <p>As a result, with <a target="_blank" href="https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html?id=us:2ps:3gl:aisgm26:awa:CONS:em:K0218784:012626:kwd-430833501819:195648817121:794247818306::&amp;gclsrc=aw.ds&amp;gad_source=1&amp;gad_campaignid=23269751971&amp;gclid=CjwKCAjw6MPRBhBTEiwAd-7Mrwla_c_gzzSu2TNPVBWdBS4INV4yNdYgtukO2OU2-zWBAuwZE-s3mBoCwegQAvD_BwE" rel="noopener">most AI projects failing</a> heading into 2026, all other data management trends have receded.</p> <section class="section main-article-chapter" data-menu-title="An array of announcements"> <h2 class="section-title"><i class="icon" data-icon="1"></i>An array of announcements</h2> <p>Models, as it turned out, weren't the only problem.</p> <p>For AI tools to be trusted, reasoning capabilities, whether provided by large language models (LLMs) such as ChatGPT and Claude or smaller, purpose-built models, need to be strong for an AI tool to respond accurately to queries. However, even as <a target="_blank" href="https://www.vellum.ai/llm-leaderboard" rel="noopener">LLM reasoning capabilities</a> close in on 100% accuracy for basic queries, most AI initiatives still stall.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> Without it, they're guessing. Guesswork doesn't work for enterprise businesses. </figure> <figcaption> <strong>Michael Bendersky</strong>Director of research, Databricks </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>Context remains the missing piece.</p> <p>"Without context, an agent is at risk of taking the wrong action, not necessarily because it malfunctioned but because it didn't have the information it needed to make the right decision," David Menninger, an analyst at ISG Software Research, told TechTarget.</p> <p>To improve <a href="https://www.techtarget.com/searchdatamanagement/post/Successful-data-analytics-starts-with-the-discovery-process">data discovery</a> and retrieval, data management vendors have centered product development efforts on enabling customers to find and feed agents the context they need.</p> <p>A week after Databricks' early January launch of Instructed Retriever, which improved on traditional <a href="https://www.techtarget.com/searchenterpriseai/definition/retrieval-augmented-generation">retrieval-augmented generation</a> (RAG) by adding parameters such as user instructions to searches to enhance retrieval accuracy, MongoDB unveiled new vector embedding and reranking models to improve the relevance of data discovered using <a href="https://www.techtarget.com/searchdatamanagement/feature/Vector-search-now-a-critical-component-of-GenAI-development">vector search</a>.</p> <p>Among many others, vendors such as Alteryx, AWS, Google Cloud, GoodData, Informatica, Snowflake, Tableau, Teradata and ThoughtSpot have followed suit.</p> <p>Not all have taken the same approach.</p> <p>Databricks aims to simplify and improve data retrieval systems. Like MongoDB, Teradata is among the vendors attempting to improve vector search to discover and operationalize context so Instructed Retriever, RAG and other retrieval pipelines can pull it in. And many are adding or augmenting <a href="https://www.techtarget.com/searchenterpriseai/tip/Advantages-of-a-semantic-layer-for-enterprise-AI">semantic modeling capabilities</a> to enable users to build context layers for AI.</p> <p>In June alone, Databricks, AWS, Microsoft and Snowflake revealed new tools that help customers create context layers for agents.</p> <p>The organizations that proactively went through the painstaking process of putting their data estates in order and building infrastructures that can handle the speed and scale of AI workloads are <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Domo-drives-customer-excitement-with-latest-AI-capabilities">putting at least some agents into production</a>.</p> <p>For example, Deloitte, which in January reported that one-quarter of organizations were able to move 40% of their AI projects into production, predicted that around half would reach that 40% mark by midyear. Similarly, Databricks' 2026 State of AI Agents report, published in January, showed a 327% increase in the number of organizations transitioning from single chatbots to multi-agent systems over the previous four months.</p> <p>"When agents are given clear, sufficient context, they can complete tasks in fewer steps and reach the right data faster, rather than spending cycles exploring dead ends," Bendersky said.</p> <p>Successes, however, still represent a minority.</p> </section> <section class="section main-article-chapter" data-menu-title="Semantic strength"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Semantic strength</h2> <p>With many enterprises aiming to build broad, <a href="https://www.techtarget.com/searchapparchitecture/tip/Multi-agent-architectures-How-coordinated-AI-systems-scale">multi-agent systems</a>, improving vector search and RAG are key to connecting agents with context. However, within this year's dominant data management trend, semantic modeling has emerged as the most common approach to improving context retrieval.</p> <p>Without semantic layers, LLM accuracy drops sharply when asked to derive outcomes from data distributed across multiple systems. But when <a href="https://www.gartner.com/en/newsroom/press-releases/2026-05-11-gartner-says-lack-of-semantics-causes-inaccurate-artificial-intelligence-agents-and-wasted-spending">semantic layers augment LLM queries</a>, they recover their accuracy.</p> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/howson_cindi.jpeg" alt="Cindi Howson, chief data and AI strategist at ThoughtSpot">Cindi Howson </div> <p>"Semantic layers are the magic," Cindi Howson, chief data and AI strategist at ThoughtSpot, told TechTarget. "[Accuracy] depends on how much context you give agents and how robust the semantic layer is, but semantic layers are the difference. The variability will be the complexity of the data model, as well as how robust that semantic and context layer is."</p> <p>Like vector embeddings, semantic modeling makes data discoverable by assigning definitions and characteristics that ensure consistency across an organization. But similar to vector search and storage, semantic modeling was, until recently, a niche capability used to augment analytics systems rather than a key component.</p> <p>Now, semantic layers are becoming so critical to agent performance that a group of data management and analytics vendors has banded together to simplify semantic modeling by <a href="https://www.techtarget.com/searchdatamanagement/news/366631576/New-consortium-to-aid-AI-by-standardizing-semantic-modeling">creating an open industry standard</a>. Without one, vendors provide proprietary semantic layers that can't interact with one another, adding complexity rather than reducing it as developers and engineers piece together AI pipelines that connect data stored in myriad systems.</p> <p>"There's the whole notion of what the data means, and we've never been very good at that," Menninger said, noting that there is no single, universal semantic language. "We had the same issue in the BI world as we do in the AI world. … We need a way to describe what the data means."</p> <p>One enterprise deploying a semantic layer to help organize its data and make it available for analytics and AI applications is supply chain management specialist Blue Yonder, which uses semantic modeling provider <a href="https://www.techtarget.com/searchdatamanagement/news/366582098/AtScale-adds-semantic-layer-support-for-AI-GenAI-models">AtScale's platform</a>.</p> <p>With At Scale's capabilities, Blue Yonder in July 2024 rearchitected its previously disparate data using semantic modeling, then connected its unified semantic layer to AI tools with AtScale's Model Context Protocol server to enable agents to call on contextually relevant data.</p> <p>"It allows us to get that clear alignment between the data and the business, and really define everything in one single place," Jeremy Arendt, senior director of analytics engineering at Blue Yonder, said at a session during AtScale's virtual Semantic Layer Summit in May. "The semantic layer became an integral part of making sure we have trust and faith in that data."</p> <p>Now, queries that previously took days to answer happen in <a href="https://www.computerweekly.com/news/366640193/Why-real-time-data-is-key-for-enterprise-AI">near real time</a>, he continued.</p> <p>"Our semantic layer is for everyone in the company who needs data," Arendt said. "We're focusing on enabling self-service [insight generation] at every layer of the business."</p> </section> <section class="section main-article-chapter" data-menu-title="Other tools of the trade"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Other tools of the trade</h2> <p>Though semantic layers have emerged as a key aspect of context retrieval for AI, they are just one part of a larger whole.</p> <p>Semantic layers map enterprise <a href="https://www.techtarget.com/searchdatamanagement/definition/schema">schemas</a> to make relevant data ready for agents. Also critical are instruction-aware retrieval layers and <a href="https://www.techtarget.com/searchsecurity/tip/How-to-implement-zero-trust-for-AI">zero-trust governance and security capabilities</a> that treat every user, device or application attempting to access an organization's data and business logic, according to Farmer.</p> <p>Data governance establishes policies and standards that determine how an enterprise's data is managed and accessed by users, including agents. Security capabilities, meanwhile, protect organizations from potential misuse by users, including agents, as well as attacks from outside threat actors.</p> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/farmer_donald.jpg" alt="Donald Farmer, founder and principal of TreeHive Strategy ">Donald Farmer </div> <p>Recently, with agents <a href="https://www.techtarget.com/searchdatamanagement/news/366644845/Agents-are-altering-data-security-needs-Oracle-responds">introducing new data security concerns</a> by increasing the scale and speed with which data is accessed and operationalized as well as enabling attackers to increase the speed and scale of threats, Oracle emphasized the need for enhanced security. However, instead of applying security measures to the operating systems historically protected by such tools, Oracle determined that security now needs to be applied at the data layer to better protect against increased threats.&nbsp;</p> <p>"Advanced retrievers treat business constraints … as strict rules rather than just suggestions for the LLM, [and] zero-trust agent security frameworks ensure that agents inherently use and cannot exceed the security clearance of human users who initialized them," Farmer said.</p> <p>Meanwhile, all the components that comprise a context retrieval system for AI -- those that foster discovery, retrieval and action -- need to be in a single environment, according to Bendersky.</p> <p>If they aren't delivered as a unified capability, connecting agents with context becomes a complex engineering exercise that has to be manually recreated for <a href="https://www.techtarget.com/searchenterpriseai/feature/Real-world-agentic-AI-examples-and-use-cases">each use case</a>.</p> <p>"Context delivery needs to be a platform capability, not an engineering project," Bendersky said. "This means access controls, lineage and auditability need to be built in… because how your agent got its context matters as much as what it said."</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> With agents requiring situational awareness to be accurate, connecting AI with relevant data and business logic has been an almost singular focus throughout 2026. https://cdn.ttgtmedia.com/rms/onlineimages/machine_learning_g1205538888.jpg https://www.techtarget.com/data-technologies/news/366646161/Data-management-vendors-race-to-connect-AI-with-context Tue, 28 Jul 2026 08:00:00 GMT <p>Two years ago, almost no states regulated AI. Today, <a href="https://labs.cloudsecurityalliance.org/research/csa-research-note-federal-ai-preemption-state-regulation-202/">more than 30 do</a>. Companies operating across multiple states often build to the strictest requirement they face, but only when the requirements differ by degree. When they differ by kind, no single bar exists. One state's rule can push a company to normalize a skewed applicant pool. Another can push the company to follow raw performance data. Compliance in one jurisdiction can create exposure in the next, forcing the enterprise to choose which law to break.</p> <p>CIOs must make AI governance decisions based on <a href="https://www.techtarget.com/searchenterpriseai/tip/Global-AI-legislation-and-regulation-tracker">today's patchwork of state laws</a>, not under the federal framework they hope Congress eventually adopts. The longer that gap persists, the more enterprise AI governance turns into legal fragmentation management rather than technology implementation. The problem lands on CIOs because AI now runs through HR, marketing, customer data, cybersecurity and legal compliance simultaneously, not just one department's tech stack.</p> <section class="section main-article-chapter" data-menu-title="The hiring challenge"> <h2 class="section-title"><i class="icon" data-icon="1"></i>The hiring challenge</h2> <p><a href="https://www.techtarget.com/searchhrsoftware/feature/AI-hiring-needs-a-decision-trail">Hiring usually shows the problem most clearly</a> because no other task or function runs AI against as many people as often. While volume doesn't make hiring the first place trouble appears, it does make hiring the place where trouble becomes hardest to miss.</p> <p>A tool that reduces 100,000 resumes to a shortlist of 10 makes judgment calls about commute times, work history and career gaps. No one wrote those judgments into policy. No one reviewed them before the shortlist reached a hiring manager. LinkedIn funnels candidates who might have never applied. Data providers scrape the open internet to identify and solicit people who never put themselves forward for anything.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> A tool trained on decades of history inherits that history's patterns. </figure> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>None of this requires bad intent. A commute-time filter can quietly exclude applicants from underserved neighborhoods without any decision-maker choosing to discriminate. An applicant pool that skews 70%-30% along gender lines raises a hard question: Should the company normalize the ratio, leave it alone or follow whatever the underlying performance data shows? The third option is the only one that looks neutral. It assumes unbiased historical performance data, which is the same assumption that fails in the retail promotion context discussed below. A tool trained on decades of history <a href="https://www.techtarget.com/searchenterpriseai/feature/The-AI-bias-playbook-Mitigation-strategies-for-CIOs">inherits that history's patterns</a>.</p> <p>There is no clean answer, and the states that have tried to write one down do more than set different bars; some point in opposite directions. The legal exposure companies face comes from disparate impact, which occurs when a neutral policy or practice disproportionately harms members of a protected group. This exposure comes from how AI performs once deployed at scale against real people, not from anyone's intent or from whether the company disclosed the practice.</p> </section> <section class="section main-article-chapter" data-menu-title="Development vs. deployment: Where's the true exposure?"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Development vs. deployment: Where's the true exposure?</h2> <p>Vendors build AI systems, but the companies using them decide where to deploy them, what role they will play in consequential decisions and whether that use complies with each state's law. The statutes don't agree on the legal hook. Some reach the developer. Some reach the deployer. Some reach both.</p> <p>The allocation shifts from one framework to the next. But one fact doesn't shift: the company operating the tool against real people remains present in every jurisdiction at once. <a href="https://www.techtarget.com/searchenterpriseai/opinion/The-legal-mistake-courts-should-avoid-in-Eightfold-AI-lawsuit">Deployment, not development, concentrates exposure</a> regardless of how any single statute assigns responsibility. A vendor's safety testing claims or federal review status doesn't move that exposure off the deployer's books.</p> <p>This dynamic extends beyond hiring. Retailers that disclose their cameras and offer an opt-out can still deliver advertising along lines that track race, gender or age. Disclosure and opt-out solve a notice problem. They do nothing about impact because the underlying model selects for engagement, not for any demographic anyone chose. The retailer can act transparently and still produce a skewed outcome. That's the point: Consent mechanisms don't cure disparate impact. Promotion decisions carry the same risk. A tool trained on decades of performance data inherits that history's patterns and can produce an outcome nobody at the company would have chosen.</p> </section> <section class="section main-article-chapter" data-menu-title="Is a single federal standard the answer?"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Is a single federal standard the answer?</h2> <p>A federal framework with one set of standards, rather than 30 separate state ones, sounds like the obvious fix. The White House has tried twice in the last year. First, it created a <a href="https://www.techtarget.com/searchenterpriseai/feature/Who-wins-and-loses-with-Trumps-AI-executive-order">task force to challenge state AI laws in court</a>. Then it created a <a href="https://www.techtarget.com/searchenterpriseai/news/366644013/Trump-AI-order-targets-frontier-model-prerelease-review">voluntary security review window for AI developers</a>. Neither substitutes for legislation. An executive order can't preempt state law without Congressional action, and a voluntary review imposes no compliance obligation on vendors that can opt in or out.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> A single federal standard remains the better end state than 30 conflicting ones. </figure> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>However, even a working mechanism wouldn't answer the harder question: What level of algorithmic bias can the law tolerate? The White House's efforts fail procedurally. The policy question remains unresolved for a different reason. Zero bias is impossible. No human being is bias-free, and neither is the data humans produce.</p> <p>No federal lawmakers, left or right, have been willing to draw that line. The problem isn't the bias itself. Anti-discrimination law has tolerated imperfect, biased human decision-making for decades without demanding zero. An algorithm changes the politics by making residual bias legible. It can measure, report and audit that bias after the fact. A statute would have to name a tolerance and defend it. That means conceding that the approved system remains biased and saying by exactly how much. Doing that is harder politically than tolerating the same bias when it spreads across a thousand human managers, and no one has to sign the number.</p> <p>A single federal standard remains the better end state than 30 conflicting ones. But one risk on the other side deserves acknowledgment, even if it doesn't change the conclusion. Technology is moving fast enough that some of today's hardest questions, especially bias normalization, may have technical answers no one has built yet. A federal standard written today could lock in today's assumptions and foreclose tomorrow's solutions. That risk calls for careful drafting. It doesn't justify the patchwork of state standards, which imposes its costs now and every day it persists.</p> </section> <section class="section main-article-chapter" data-menu-title="What to do today"> <h2 class="section-title"><i class="icon" data-icon="1"></i>What to do today</h2> <p>None of this changes <a href="https://www.techtarget.com/searchenterpriseai/feature/AI-regulation-What-businesses-need-to-know">what companies can do today</a>. CIOs should map AI by use case and jurisdiction, so they know which laws govern each deployment. They should then build governance around practices that should exist regardless of the regulatory landscape, including clear terms of service, reliable data provenance and human review where the consequences matter most.</p> <p>Congress, courts and future administrations will keep shaping AI regulation, but enterprise AI won't wait for them. Companies already deploy systems under laws that states can enforce today. CIOs must govern within the legal landscape that exists now, not the federal framework they hope will eventually arrive.</p> <p><i>Jon Polenberg is a shareholder and vice chair of Becker's Business Litigation Practice. As a business trial lawyer in Florida, he's known for his advocacy and strategic insight in complex commercial litigation. Jon has decades of experience representing businesses in high-stakes cases and handling intricate legal disputes with precision and a commitment to excellence. His practice spans a range of industries, assisting companies in resolving matters that affect their operations, reputations and financial interests.</i></p> </section> Companies are deploying AI across their organizations faster than lawmakers are building consistent rules to govern it, leaving CIOs to navigate an expanding patchwork of state requirements. https://cdn.ttgtmedia.com/rms/onlineimages/ai_a264431831.jpg https://www.techtarget.com/ai/opinion/How-to-manage-the-gap-between-enterprise-AI-use-and-AI-regulation Mon, 27 Jul 2026 13:35:00 GMT <p>Context makes or breaks an AI agent.</p> <p>It's what gives AI agents the situational awareness to deliver trustworthy, accurate outputs. Without it, all other parts might be in place, but the relevant meaning that powers an agent to perform properly will be absent, dooming it to fail.</p> <p>When provided with proper awareness, agents can <a href="https://www.techtarget.com/healthtechanalytics/feature/LLMs-struggle-with-clinical-reasoning-study-finds">help make medical diagnoses</a>, personalize retail experiences, discover new clients, generate sales campaigns, manage and optimize entire supply chains and monitor vast swaths of data that are impossible for humans to oversee to detect fraud. Sometimes, they're even building other agents.</p> <p>But to deliver those outputs, agents need proper context. They need relevant, high-quality data and task-appropriate <a href="https://www.techtarget.com/whatis/definition/business-logic">business logic</a>, and the ability to discover and ingest them independently.</p> <p>"It's no different than providing context to an employee doing a job," David Menninger, an analyst at ISG Software Research, told TechTarget. "You wouldn't give a junior analyst a task without giving them some context."</p> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/menninger_david.jpg" alt="ISG Software Research analyst David Menninger">David Menninger </div> <p>Now, after the laborious task of getting data ready for AI and investing heavily in experimental pilots -- while model reasoning capabilities improved -- some enterprises are ready to put agents into production at scale and reap the benefits.</p> <p>The most advanced organizations already have. But widespread deployment of <a href="https://www.techtarget.com/searchapparchitecture/tip/Multi-agent-architectures-How-coordinated-AI-systems-scale">multi-agent networks</a> requires difficult work that could take years to complete, according to Menninger.</p> <p>"We'll … gradually eat away at the parts that are difficult, and we'll hopefully, in a several year time period, have tackled the biggest challenges," he said.</p> <section class="section main-article-chapter" data-menu-title="AI success amid failures"> <h2 class="section-title"><i class="icon" data-icon="1"></i>AI success amid failures</h2> <p>Despite the time and money that enterprises have invested in agentic AI development, most AI projects don't make it into production.</p> <p>Studies show that the failure rate is going down, which is evidence that enterprises that have addressed the underlying data issues that often stall AI projects are achieving at least some success connecting AI applications with the context they need to deliver trustworthy outputs.</p> <p>But they also reveal that more AI projects still fail than succeed.</p> <blockquote class="main-article-pullquote"> <div class="main-article-pullquote-inner"> <figure> It's no different than providing context to an employee doing a job. You wouldn't give a junior analyst a task without giving them some context. </figure> <figcaption> <strong>David Menninger</strong>Analyst, ISG Software Research </figcaption> <i class="icon" data-icon="z"></i> </div> </blockquote> <p>For example, Deloitte's 2026 <a target="_blank" href="https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html?id=us:2ps:3gl:aisgm26:awa:CONS:em:K0218784:012626:kwd-430833501819:195648817121:794247818306::&amp;gclsrc=aw.ds&amp;gad_source=1&amp;gad_campaignid=23269751971&amp;gclid=CjwKCAjw6MPRBhBTEiwAd-7Mrwla_c_gzzSu2TNPVBWdBS4INV4yNdYgtukO2OU2-zWBAuwZE-s3mBoCwegQAvD_BwE" rel="noopener">State of AI in the Enterprise report</a>, released in January, showed only one-quarter of the organizations surveyed have been able to move 40% of their AI experiments -- whether single agents, multi-agent systems, chatbots, traditional machine learning projects or other pilots -- into production. Similarly, Menninger noted that ISG's year-end 2025 research found that about one-third of the 1,200 projects it examined are moving into production, an improvement from 17% a year earlier but still well under 50%.</p> <p>"[Some] enterprises are no longer experimenting with agents," Michael Bendersky, director of research at Databricks, told TechTarget. "They're deploying them for real, complex work."</p> <p>Databricks' own <a target="_blank" href="https://www.databricks.com/resources/ebook/state-of-ai-agents" rel="noopener">2026 State of Agents report</a>, published in January, found a 327% increase in usage of multi-agent systems built on domain intelligence over the previous four months, demonstrating that some enterprises -- after there were almost none until late 2025 -- have built at least remedial agentic systems.</p> <p>Nevertheless, long-term challenges to building broad networks of context-aware agents remain.</p> </section> <section class="section main-article-chapter" data-menu-title="Progress despite problems"> <h2 class="section-title"><i class="icon" data-icon="1"></i>Progress despite problems</h2> <p>One challenge to putting multi-agent systems into production is the completeness and quality of the AI workflow, according to Bendersky.</p> <p>Humans used to perform all aspects of <a href="https://www.techtarget.com/searchbusinessanalytics/feature/Top-data-preparation-challenges-and-how-to-overcome-them">data preparation</a>, retrieval and operationalization for analytics and AI, and they had time to ensure the quality of all parts of the pipelines that connect data with applications. Once they're in production, agents are the ones calling on context, and for them to do so properly, everything they require for accuracy must be ready the instant they need it.</p> <p>Enterprises that haven't already done so need to modernize -- and in some cases, overhaul -- their data and AI infrastructures, which can be <a href="https://www.techtarget.com/searchenterpriseai/feature/The-dollars-and-sense-of-implementing-AI">time-consuming and costly</a> depending on how much needs to be revamped.</p> <p>To connect agents with context, infrastructures need tools that improve data discovery to ensure agents call on context-appropriate information. Among others, they include the following:</p> <ul class="default-list"> <li>Vector embedding and reranking models.</li> <li>Semantic layers and <a href="https://www.techtarget.com/searchdatamanagement/feature/Hybrid-search-demands-reshape-retrieval-frameworks-for-AI">data retrieval</a> engines that are purpose-built for AI workloads rather than BI tasks.</li> <li>Security capabilities that address new risks posed by agents.</li> <li>Governance frameworks that oversee not only data but also agent behavior.</li> </ul> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/bendersky_michael.jpg" alt="Michael Bendersky, director of research at Databricks">Michael Bendersky </div> <p>"What will still remain difficult is the environment," Bendersky said. "We are moving from an enterprise where most data access is done by humans to one where interactions with the data are done by agents. This will require continued investment in architectures where the right context is accessible to the agent at the right time. Governance, platforms, auditability, etc., will continue being challenging."</p> <p>In particular, <a href="https://www.techtarget.com/searchdatamanagement/opinion/Why-data-semantics-matters-for-context-aware-systems">semantic modeling capabilities</a> are proving to be a crucial enabler for enterprises successfully building agents and need to be part of an organization's AI workflow, according to Cindi Howson, chief data and AI strategist at ThoughtSpot.</p> <p>Model reasoning capabilities have improved to the point that when large language models (LLMs) are asked questions that require them to call on limited amounts of contained data, they are 95% accurate. However, <a target="_blank" href="https://axiussdc.substack.com/p/the-7-table-fallacy-why-text-to-sql?utm_campaign=post&amp;utm_medium=web&amp;triedRedirect=true" rel="noopener">a September 2025 study</a> demonstrated that when LLMs are asked to derive outcomes based on broad swaths of data spread across myriad systems, their accuracy falls to around 50% or lower.</p> <p>A separate <a target="_blank" href="https://docs.getdbt.com/blog/semantic-layer-vs-text-to-sql-2026?version=2.0&amp;name=Fusion" rel="noopener">April 2026 study</a> by DBT Labs -- which provides semantic modeling capabilities -- showed that accuracy again approaches 100% when LLM queries, often using traditional <a href="https://www.techtarget.com/searchenterpriseai/definition/retrieval-augmented-generation">retrieval-augmented generation</a> (RAG) pipelines to connect models with AI tools, are augmented by a semantic layer.</p> <p>"Some organizations thought they could just use a basic RAG approach, and it's not happening -- it's just not good enough," Howson told TechTarget. "[They need] semantic layers and context layers, and humans in the loop to coach agents."</p> </section> <section class="section main-article-chapter" data-menu-title="The next challenge(s)"> <h2 class="section-title"><i class="icon" data-icon="1"></i>The next challenge(s)</h2> <p>While known problems slowly get solved, new ones constantly emerge.</p> <p>For example, connecting agents with context only became the dominant trend in data management and AI at the start of 2026 after unreliable model reasoning was viewed as the initial culprit holding back agentic AI development. Only once <a target="_blank" href="https://www.vellum.ai/llm-leaderboard" rel="noopener">model reasoning capabilities</a> reached a certain threshold and agents still weren't consistently delivering accurate outputs did context become the primary focus, with a litany of vendors introducing tools to connect agents and context.</p> <p>Now, data and AI infrastructures that don't effectively connect agents with context are a principal focus, but as Menninger noted, it will take time to solve on a widespread basis. In the interim, Incremental improvements will be made that enable more enterprises than those now at the forefront to move agents into production and perhaps even build multi-agent networks.</p> <p>"We'll be making progress, identifying the bigger challenges and devoting research to solving those," Menninger said.</p> <p>But what then?</p> <p>Assuming a combination of enterprise investment and technological advancement continues to result in a rising AI development success rate, <a href="https://www.techtarget.com/searchapparchitecture/tip/Multi-agent-architectures-How-coordinated-AI-systems-scale">managing systems</a> that include thousands of autonomous agents will be a significant challenge, according to Donald Farmer, founder and principal of TreeHive Strategy.</p> <div class="imagecaption alignLeft"> <img src="https://cdn.ttgtmedia.com/rms/onlineimages/farmer_donald.jpg" alt="Donald Farmer, founder and principal of TreeHive Strategy">Donald Farmer </div> <p>"Multi-agent systems -- or swarms – [are one emerging issue]," he told TechTarget.</p> <p>Agents within connected systems are meant to work together, to collaborate on a scale beyond human capacity. But for them to do so without overstepping boundaries, without exposing sensitive information, without accessing data they're not authorized to operationalize, without violating regulatory statutes, without breaching <a href="https://www.techtarget.com/whatis/definition/data-sovereignty">data sovereignty</a> rules -- without behaving exactly as intended -- is challenging.&nbsp;</p> <p>"Once individual agents have great context, [organizations have to figure out] how do you get an accounting agent, a legal agent and a procurement agent to collaborate, negotiate and hand off tasks to one another without human intervention or emergent anomalies," Farmer said.</p> <p>Another issue enterprises will have to address is how much responsibility to give agents and how much to keep under human supervision, according to Menninger.</p> <p>He noted that ISG research shows that of those agents currently in production, 50% assist humans, 40% work in conjunction with humans but will eventually be empowered to act autonomously, and 10% are fully autonomous. Figuring out <a href="https://www.techtarget.com/searchdatacenter/tip/AI-operating-models-Balancing-autonomy-and-human-oversight">the right balance</a> will be critical for enterprises to derive the benefits of agentic AI without suffering accidental consequences.</p> <p>"There's always going to be a next problem or opportunity," Menninger said.</p> <p><i>Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.</i></p> </section> Awareness enabled by appropriate data and business logic is one of the differences between agentic applications that properly perform and those that never reach production. https://cdn.ttgtmedia.com/rms/onlineimages/machine%20learning_g1186820873.jpg https://www.techtarget.com/data-technologies/news/366646252/Context-is-king-as-agents-evolve-but-problems-persist Mon, 27 Jul 2026 08:00:00 GMT 60 webmaster@techtarget.com