Oracle AI Database update aims to ease developing agents

Fueling agents

Worldwide, investments in AI have surged since OpenAI's November 2022 launch of ChatGPT marked significant improvement in generative AI technology and showed how GenAI can make business employees better informed and operations more efficient.

Oracle is adding AI to the database, but more importantly brings unstructured information [so] vectors can be called and used with transactional data. It's unique and powerful approach. CIOs love it as they do not have to move -- and then worry -- about their transactional data.

Research and advisory firm Gartner predicts that spending on AI initiatives will total $25 trillion in 2026 and $3.3 trillion in 2027, up from $1.75 trillion in 2025 and just under $1 trillion in 2024.

However, despite increasing their investments in AI projects, most enterprises have not been able to move pilots into production to derive benefits from AI tools, with MIT surmising that 95% or organizations have gotten zero return on their investments.

While there are many reasons AI initiatives fail at such a high rate, poor data retrieval and security concerns are among them.

Oracle's AI Database update aims to better enable users to develop AI applications by improving their access to the relevant data needed to feed AI tools so they deliver appropriate outputs. In addition, the update is designed to make it more secure to operationalize data for AI.

"Organizations are increasingly recognizing that embedding AI capabilities directly alongside their critical business data significantly enhances performance and accelerates the deployment of AI agents," said Maria Colgan, Oracle's vice president of product management for AI and mission-critical data. "This served as the key driver behind this architectural decision."

In particular, embedding AI next to proprietary data within a user's database eliminates the need to send data across systems, which improves both accuracy and security, she continued.  

"That means its answers are more accurate and grounded in what's happening right now -- not a stale copy from somewhere else," Colgan said. "Just as importantly, everything stays under the database's built-in security controls. The agent only sees what it's supposed to see -- nothing more -- so you're not exposing extra data just to make the system work."

Oracle's AI Database update includes the following new capabilities:

• Oracle Autonomous AI Vector Database, which features APIs and an easy-to-use web interface so that developers, data scientists and other users can easily index data to make it discoverable so it can be used to inform vector-powered applications.
• Oracle AI Database Private Agent factory, a no-code tool for developing agents that includes a Database Knowledge Agent, a Structured Data Analysis Agent and a Deep Data Research Agent to help users safely and securely create and deploy cutting-edge applications.
• Oracle Unified Memory Core, a low-latency reasoning engine that enables users to store vector, JSON, graph, relational, text, spatial and columnar data in one system so it can be discovered for AI development.
• Oracle Deep Data Security to deploy data privacy rules specific to each user or agent that has access to ensure proper data security.
• Oracle Private AI Services Container so that customers with heightened security requirements can run private instances of AI models and avoid sharing data outside their secure environments, including with third-party AI providers.
• Oracle Trusted Answer Search, a feature that uses AI vector search rather than large language models to respond to queries and provides testable outputs to avoid hallucinations and misunderstandings.

Like Mueller, Stephen Catanzano, an analyst at Omdia, a division of Informa TechTarget noted the importance of Oracle's new AI Database capabilities given that they enable users to integrate AI with proprietary data across both operational databases and lakehouses.

"The new Oracle AI Database capabilities are significant additions as they … eliminate the need for complex data pipelines, enhance security and allow users to activate their data for AI-driven insights and applications, which were previously challenging to achieve at scale," he said. "This tighter integration is a trend I see with most vendors now."

In addition, the update could give Oracle a slight advantage over platforms such as Microsoft Fabric and Google BigQuery as well database capabilities from specialists including MongoDB, Catanzano, who named Private Agent Factory the most valuable of the new features, continued.

"The new features, such as unified memory core and deep data security, help differentiate Oracle by offering converged data processing and robust security measures tailored for AI workloads, though other vendors are also advancing similar capabilities," he said.

Looking ahead

As Oracle plans future AI Database updates, adding more AI capabilities is a priority, according to Colgan.

With agents already part of Oracle AI Database, and with new database capabilities designed to help customers more effectively build their own agents, Oracle would be wise to add capabilities that enable customers to monitor agents, according to Mueller.

"Oracle, with all this, has passed the first hurdle, [which] is the data phase for AI," he said. "Then comes the management of agents. One step after another."

Catanzano, meanwhile, suggested that Oracle could better serve users and perhaps attract new ones by further expanding the capabilities of its AI Database as well as integrating AI tools from third parties.

"Oracle could further enhance its AI Database by integrating advanced generative AI capabilities for predictive analytics and decision-making, as well as expanding support for emerging AI frameworks," he said. "This would not only address evolving user needs but also attract new customers seeking cutting-edge AI solutions."

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.