IBM delivers AI agent orchestration innovations at Think 2025
IBM's focus on AI agent orchestration, observability, and governance supports enterprise IT departments' need for a tool that spans multiple vendor platforms.
- Mark Beccue,
-
Enterprise Strategy Group
We provide market insights, research and advisory, and technical validations for tech buyers.
Significant trends are emerging for enterprise AI: market demand and investment in AI agents and widespread efforts to drive down the cost of AI workloads.
Despite commercial products emerging just within the last year, the AI agent market is experiencing rapid growth. Unlike broader generative AI discussions that often veer into speculative territory, AI agents represent more practical applications focused specifically on business efficiency, workflow automation, and enhanced decision-making capabilities.
Organizations are prioritizing AI agents over other AI initiatives, reflecting their confidence in potential ROI through tangible productivity and efficiency gains. However, our research shows that organizations face significant challenges in their AI adoption journeys, particularly around system interoperability and security. As organizations increasingly deploy AI agents that need to interact with external systems and other agents, issues around permissions, security protocols, and cross-system communication will become more prominent concerns that must be addressed.
Running parallel to the AI agents trend, enterprises are becoming increasingly focused on understanding and reducing the total cost of ownership (TCO) of AI. According to recent research on AI infrastructure trends by Enterprise Strategy Group, now part of Omdia, 55% of organizations are planning to spend $2.5 million or more on AI initiatives over the next 24 months, and 42% say cost efficiency is one of the most important attributes for their AI infrastructure stack. Taking it a step further, the research showed that the top challenge organizations face in operationalizing AI is the high costs associated with implementation.
AI agent orchestration across vendor platforms
In a May 6, 2025 blog post, IBM's General Manager for Data and AI Ritika Gunnar had this to say about AI agents and how IBM is working with enterprises to make them a reality: "…agents must be able to work seamlessly across the vast web of applications, data and systems that underpin today's complex enterprise technology stacks. Which means that orchestration, integration and automation are the secret weapons that will move agents from novelty into operation."
With that, IBM introduced a set of agent capabilities to help companies build, run and manage agents, delivered via IBM watsonx Orchestrate.
The prebuilt agents are designed for specific business domains and use cases, with ready-to-use skills and integrations. They target specialized areas including HR, sales, and procurement, with plans for additional domains in the coming months.
IBM also launched a simplified process for building, customizing and deploying agents for companies that want to build their own. According to IBM, the no-code tool option allows users to build an agent in five minutes, or developers can use pro-code tools.
With agentic orchestration, users can integrate and automate agents to take on complex projects. AI agents and assistants can work together across any tool, data source, or infrastructure. In addition, agent observability can be used to discover, monitor and optimize the use of agents across the enterprise to drive trust, performance and efficiency.
All in all, IBM watsonx Orchestrate functions as a multi-agent supervisor, router, and planner, enabling agents, whether from IBM, third-party platforms or open source, to collaborate and operate with a company's existing assets, which may include traditional automation software, AI assistants, APIs, data stores, and various business applications. The platform is preintegrated with over 80 enterprise applications, "making it the ideal gateway for infusing agents across the full enterprise stack," Gunnar said in the blog post.
AI agent ecosystem, infrastructure
At Think 2025, IBM positioned itself as the pragmatic AI leader, outlining key initiatives for both AI agents and AI infrastructure TCO.
IBM's approach to AI agents through watsonx Orchestrate represents a significant strategic move within the AI agent ecosystem, particularly through direct integrations with seven major enterprise software players, including Salesforce, Microsoft, Adobe, ServiceNow, and Oracle.
The integration strategy differentiates IBM from competitors by enabling seamless multi-agent and cross-software tasks across major enterprise platforms. The company's focus on AI agent orchestration, observability, and governance demonstrates a mature understanding of enterprise needs. This isn't just another API play; rather, it's real, native integrations that let AI agents work across these major platforms seamlessly. That's a big deal for enterprises looking to automate complex processes.
What's particularly striking is the scale of IBM's integration network: It has 80 partners total, with those seven key players leading the charge. This kind of ecosystem play puts IBM in a different league compared to what others are doing with agent-to-agent protocol.
In addition to agentic AI, IBM LinuxOne Emperor 5 has the potential to significantly reduce the cost of AI inferencing. The new system is engineered to help address cost, space, and energy challenges, said IBM Z and LinuxOne CTO Marcel Mitran and Chief Product Officer Tina Tarquinio, in a blog post.
"For example, organizations can reduce operational complexity and optimize their IT infrastructure by consolidating workloads across multiple servers onto a single high-capacity system. Moving cloud-native, containerized workloads from a comparable x86 solution to an IBM LinuxONE 5 running the same software products can save up to 44% on the total cost of ownership over 5 years," they said in the blog post.
The LinuxOne Emperor 5 numbers were eye-opening, especially on the TCO when compared to x86 systems. If IBM's hardware efficiency stats hold up, we will see more companies bring AI inferencing back to on-premises mainframes because the economics make sense. It's another indication of the momentum and preference enterprises are forming to run their AI workloads outside of the public cloud.
Enterprise compute is not public cloud or on-premises; it's both. This reality supports IBM's hybrid cloud philosophy. Its Salesforce partnership is a perfect example, as the partnership makes it possible for customers to tap into mainframe data for cloud AI applications.
This is exactly what enterprises have been asking for. This approach acknowledges where most enterprises actually are -- sitting on mountains of valuable data in their mainframes and legacy systems but wanting to use it for modern AI applications.
I was impressed with IBM Infrastructure CTO Hillery Hunter's keynote. Instead of just pushing shiny new tools, she laid out a practical vision for getting AI infrastructure right. This resonates strongly with what we're seeing in the market: our research shows that 33% of organizations consider TCO and ROI essential for an enterprise-ready AI system. IBM seems to get that organizations need to nail the fundamentals to scale AI.
IBM is showing its thinking several moves ahead in the enterprise AI game by addressing real implementation challenges and infrastructure needs. The company's focus on practical, enterprise tools and strong partnerships could give it a serious edge in the AI race. While others are still talking about what's possible, IBM is actually building the bridges to make it happen.
Mark Beccue is principal analyst at Enterprise Strategy Group, now part of Omdia, covering artificial intelligence.
Enterprise Strategy Group, now part of Omdia, analysts have business relationships with technology vendors.
