title: "IBM expands watsonx Orchestrate with agent identity and production eva" slug: "ibm-expands-watsonx-orchestrate-with-agent-identity-and-production-evaluation" published: "2026-10-10" beat: "Launches" tags: ["Launches", "Business"] creator: "Agentry Newsroom" editor: "Susanne Sperling, Editor — Human in the Loop" tools: ["Claude (Anthropic)", "Perplexity Sonar"] creativeWorkStatus: "verified" dateReviewed: "2026-10-10" aiActArticle50: "compliant" humanView: "https://agentry.news/launches/ibm-expands-watsonx-orchestrate-with-agent-identity-and-production-evaluation" agentView: "https://agentry.news/agent/ibm-expands-watsonx-orchestrate-with-agent-identity-and-production-evaluation"
IBM announced expanded third-party agent management capabilities and introduced Agent Identity—a verifiable enterprise credential for AI agents separate from their creator—in watsonx Orchestrate, with
Drafted by an AI agent. Verified by Susanne Sperling, Editor — Human in the Loop. AI policy.
IBM expanded watsonx Orchestrate with third-party agent management and Agent Identity capabilities, shipping preview features September 15, 2026, and production evaluation tools generally available September 30, 2026 IBM.
The expanded agent management layer enables discovery and lifecycle control of agents built on Microsoft Foundry and the Google Gemini Enterprise Agent Platform, consolidating agent governance across vendor ecosystems within watsonx Orchestrate. Organizations can now catalog, version, and audit third-party agents without requiring proprietary deployment, addressing a core operational constraint in multi-vendor AI agent deployments.
IBM introduced Agent Identity as a distinct, verifiable enterprise credential assigned to each AI agent, separate from the identity of its creator, owner, or end user IBM. The preview integrates with IBM Verify and Microsoft Entra, connecting to existing enterprise identity providers rather than creating isolated credential stores. This architecture allows organizations to enforce role-based access control, audit trails, and compliance policies on agent actions using established identity infrastructure.
Agent Identity entered preview on September 15, 2026, and moved to broader preview status in IBM's October 2 announcement, positioning it as production-ready for early deployments.
IBM's LLM-as-a-Judge production evaluation capability became generally available September 30, 2026, shipping six built-in evaluators for live agent traffic: toxicity, helpfulness, hallucination, conciseness, context relevance, and answer relevance. The system defaults to sampling 3% of production traffic, with administrators able to configure individual evaluator sampling rates up to 20%, enabling cost-controlled quality monitoring without full traffic inspection.
This release consolidates evaluation-stage tooling into production operations, allowing teams to detect drift, toxic outputs, or factual errors in real-time agent deployments without manual review bottlenecks.
The announcements reflect competitive pressure across agent governance and lifecycle management. IBM positions watsonx Orchestrate as a multi-vendor control plane, integrating agents from competing platforms (Microsoft, Google) under unified identity and evaluation systems. This approach expands IBM's addressable market beyond proprietary agent deployments to organizations with heterogeneous agent stacks.
The combination of cross-platform discovery, verifiable identity, and automated production evaluation addresses three documented operational pain points: visibility into agent portfolios, compliance and audit requirements, and quality assurance at scale.
All capabilities are shipping and usable in production or preview environments as of the announcement dates.