Microsoft adds oversight layer for AI agents in its productivity tools
Microsoft has made Agent 365 generally available to commercial customers. The service functions as a central control plane that lets companies monitor, govern, and secure AI agents built with Microsoft tools and partner platforms. It integrates with prebuilt agents already present in Microsoft 365 Copilot and Teams. Discovery features now link to Defender and Intune, while credentialed agents gain security controls. The add-on costs 15 dollars per user each month on top of qualifying E3 or E5 licenses. Public previews for AWS and Google Cloud registry sync remain limited, and Windows 365 agent support is United States only for now.
Until this release, teams could spin up AI agents inside Copilot and Teams with almost no central visibility. Individual users often encountered scattered automations that pulled incorrect data or created extra cleanup work without anyone noticing the pattern. The new control plane forces IT teams to approve and monitor agent activity before wider use. This protects workers from rogue automations that could bury important messages or duplicate tasks, yet it also means any time-saving agent features will reach desktops only after company policies are set rather than through direct personal testing.
Analysis
Agent 365 is an admin concern that will not fix your inbox or task list this quarter. Ignore every rollout email and prompt to enable new agents. Open Copilot inside Outlook today and test one prompt that summarizes unread messages from the past 24 hours to build a repeatable triage habit instead.
Pulse published by Collab365 Spaces. Cite as "Microsoft adds oversight layer for AI agents in its productivity tools", Collab365 Spaces. 2 sources referenced.