Microsoft Launched the First Enterprise Control Plane for AI Agents

Microsoft Agent 365 became generally available on May 1, alongside the new Microsoft 365 E7 suite. It is the first centralized control plane designed to discover, govern, and secure every AI agent operating in an enterprise environment, whether built by Microsoft, third-party vendors, or employees themselves. (Microsoft Security Blog | Microsoft Learn | Redmond Magazine)

The product solves a problem most enterprises did not realize they had until recently: shadow agents. Through Defender and Intune integration, Agent 365 can detect AI agents running on Windows devices, including OpenClaw agents, GitHub Copilot CLI, and Claude Code, and enforce policies through the Intune admin center. IT teams get a registry of every agent in their environment, with the ability to block, restrict, or audit agent activity at the device and identity level.

The architecture rests on three pillars: observe (discover and inventory agents across the tenant), govern (apply identity, access, and data-loss-prevention policies), and secure (extend Microsoft Defender protections to agent-initiated actions). Pre-integrated ecosystem partner agents are available to deploy directly from the Microsoft 365 admin center. Microsoft recommends Entra P1/P2 and Purview Data Loss Prevention for full capability.

The timing matters. This launches the same week the Five Eyes published agentic AI security guidance (we covered this yesterday) and three days after the PocketOS incident demonstrated what happens when agents operate without governance infrastructure. Microsoft is betting that "agent governance" becomes a procurement category as fundamental as endpoint management. Given the trajectory of enterprise AI adoption (79% of companies have adopted agents, per recent surveys, but only 2% have fully deployed them), the gap between adoption and governance is exactly where Agent 365 is positioned.

Roadmap signal: If you are building agent platforms that operate in enterprise environments, Agent 365 sets the expectation that agents will be subject to the same identity, access, and compliance controls as human users. Design for discoverability, auditability, and policy enforcement from the start, or expect Microsoft's control plane to enforce those constraints for you.