Introduced identity, access-control, and monitoring capabilities to help enterprises manage AI agents across business systems. (Oct. 27)
On Oct. 27, 2026, at TechXchange, IBM announced new AI-agent controls spanning IBM Agent Identity, HashiCorp Vault and watsonx Orchestrate. Agent Identity records an agentβs owner, lifecycle status and delegated authority; Vault supplies short-lived credentials and least-privilege access, with real-time denial and session revocation if an agent exceeds its permissions. Orchestrate adds cross-platform discovery, isolated sandboxes, pre-deployment testing and production monitoring, including token-use and cost tracking; it can find agents built with Amazon Bedrock, Microsoft Foundry and Google Gemini. IBM also described testing agents, models and MCP servers for quality, accuracy and compliance. The move addresses a growing enterprise problem: agents can act across business systems, but security teams need to know who authorized them, what they can access and what they did. It positions IBM to manage agents across competing cloud ecosystems rather than only its own, while giving IT and security teams a unified control point.