Microsoft adds centralized agent registry to Agent 365

Agent 365 already existed as Microsoft's administrative offering, and this update is Microsoft's answer to a problem it says has grown as organizations deploy AI agents across different environments. Because agents built in different platforms can run with varying permissions, data access levels, and compliance controls, IT teams often lack a centralized way to track what is running or what data it can reach. Microsoft compares unmanaged agents to shadow IT, warning that this fragmentation can increase security gaps, policy violations, and unchecked growth.

The registry itself is a synchronization and management layer rather than a new agent-building tool. Admins can pull agents in from Amazon Bedrock, Google Vertex AI, Salesforce Agentforce, and Databricks and see them next to Microsoft's own, with monitoring, approval workflows, policy enforcement, and lifecycle management applied on top. The practical effect is that governance is meant to travel with the agent across platforms, instead of being managed separately in each environment.

Whether that trade-off pays off is likely to hinge on how much setup work the integrations demand in practice. Microsoft itself notes that the controls introduce additional administrative oversight, and that the centralized approach may slow the rapid agent deployment some IT teams are used to. For administrators weighing governance against speed, the test will be whether the registry's visibility justifies the approval workflows and policy maintenance it requires.