At Think 2025,
IBM sharpened its focus on enterprise-grade agentic AI by announcing a series of deep integrations with cloud giants Oracle, AWS, and Salesforce. These partnerships are more than technical collaborations and signal a broader shift toward enabling intelligent agents that can operate autonomously across hybrid and multi-cloud environments, all while meeting enterprise standards for compliance, scalability, and data control.
Each alliance brings IBM’s watsonx Orchestrate platform and Granite AI models closer to the operational fabric of modern enterprises, enabling context-aware automation of routine and complex workflows alike.
With Oracle, IBM is deploying its agentic AI stack—including
watsonx Orchestrate and Granite—on Red Hat OpenShift within Oracle Cloud Infrastructure (OCI). This allows enterprises to run AI-powered HR, procurement, and sales processes autonomously in whichever cloud environment best fits their regulatory and architectural needs—be it public, private, or sovereign.
On AWS, IBM has connected watsonx Orchestrate to Amazon Bedrock and Amazon Q, unlocking the ability for agents to act on real-time data pulled from widely used enterprise platforms like Salesforce, Slack, and Zendesk. watsonx.governance ensures these interactions remain secure and auditable, with governance tools readily accessible through the AWS Marketplace.
IBM’s Salesforce integration, meanwhile, focuses on bridging legacy and modern data systems. Using a Zero Copy architecture, data from IBM Z and Db2 mainframes can now flow directly into Salesforce Data Cloud—enabling AI agents to support tasks such as lead qualification, customer engagement, and internal support, all from within Salesforce and Slack environments.
Enter Agent Connect: IBM’s Ecosystem for AI Agent Builders
To complement these cloud-scale deployments, IBM introduced
Agent Connect, a new technical framework and partner program designed to help software vendors, developers, and enterprises build, deploy, and orchestrate AI agents within the
watsonx Orchestrate ecosystem.
At its core, Agent Connect provides a foundation for seamless multi-agent collaboration. It addresses the fragmentation challenge many businesses face today—where different teams experiment with disconnected agents, each running on siloed stacks. Agent Connect allows these agents, regardless of origin or framework, to work together under a unified governance model.
“IBM offers a range of resources for partners to easily integrate their agents with Agent Connect,” said
Suzanne Livingston, Vice President, watsonx Orchestrate Agent Domains at IBM. “These include robust documentation and a self-service portal with rich examples, an
Agent Development Kit offering a
pro-code environment, and an active IBM Community forum for technical support.”
The platform is designed to be framework-agnostic, supporting a wide array of agent development tools and runtimes. “Enterprises want choice and control when deploying agents,” Livingston added. “That’s why Agent Connect supports LangChain, LangGraph, CrewAI, Copilot Studio, and even custom frameworks. Participants can bring their own LLMs, including open-source models. We’ve made sure the flexibility is built in.”
Tools, Catalog, and Monetization for Builders
Agent Connect isn’t just about integration; it is also about enabling innovation and monetization. Partners can build, test, and optimize agents directly within watsonx Orchestrate, using the same environment that powers production workflows. “Through Agent Connect, partners gain access to a fully supported development and test environment,” Livingston said. “They can publish and iterate directly through the Agent Catalog.”
On the other hand, the Agent Catalog functions as a centralized marketplace where IBM-built, third-party, and open-source agents can be discovered and deployed by enterprises. It also acts as a visibility engine. “Partners gain visibility through IBM’s broader marketing engine, events, blogs, and exposure to our enterprise clients using watsonx Orchestrate,” said Livingston. Startups like Brighthive and symplistic.ai are already preparing to launch their agents alongside offerings from established partners like Box, Oracle, and Salesforce.
Monetization is a key component. “Partner agents published in the Agent Catalog can be made available for sale directly by IBM sellers to IBM’s enterprise install base,” Livingston noted. This provides smaller partners with go-to-market acceleration while giving customers a trusted channel for AI solutions.
Avoiding the Agent Sprawl Trap
As enterprises race to adopt AI-powered tools, a common risk is emerging: agent sprawl. Teams may deploy point solutions in silos, leading to a tangled mess of disconnected bots and assistants that are hard to govern and scale.
IBM is positioning Agent Connect and watsonx Orchestrate as a response to that problem. By providing a governed, interoperable ecosystem for agents—one where flexibility doesn’t come at the cost of oversight—IBM aims to help enterprises build cohesive AI strategies that grow with them.
Instead of managing dozens of isolated tools, businesses can operate a network of intelligent agents that share context, coordinate actions, and deliver measurable value across functions. For organizations seeking to operationalize AI without the complexity, IBM’s agentic approach offers a clear and structured path forward.
