Microsoft just launched Agent 365, a governance platform that lets IT teams track, approve, and shut down AI agents across their organizations. The move comes as companies struggle with an explosion of unmanaged AI tools that pose security risks and compliance headaches. Microsoft's timing couldn't be better - enterprises are deploying AI agents faster than they can monitor them.
Microsoft is tackling one of enterprise AI's biggest headaches: the complete lack of visibility into what artificial intelligence agents are actually running inside corporate networks. The company's new Agent 365 platform, announced at this week's Ignite conference in San Francisco, gives IT administrators something they've been desperately missing - a central dashboard to see, approve, and kill AI agents before they cause problems.
"In the same way you provision an identity for a new employee or a contingent worker, you'll provision identity and access controls for your agents," Judson Althoff, CEO of Microsoft's commercial business, told CNBC in an interview. It's a telling comparison - Microsoft is essentially treating AI agents like employees who need proper onboarding and oversight.
The platform automatically discovers agents from major enterprise software vendors including Adobe, Cognition, Databricks, Glean, ServiceNow, and Workday. It also tracks homegrown agents built with Microsoft's own Azure AI Foundry and Copilot Studio. More importantly, it can spot security risks and shut down problematic agents with a few clicks.
EY, one of the world's largest accounting firms, had already built an internal catalog of its AI agents but is now implementing Agent 365 to gain better control, according to Mark Luquire, a managing director at the company. The fact that even sophisticated organizations like EY need help managing their AI agent sprawl shows how quickly this problem has escalated.
The timing reflects a broader enterprise reality: companies jumped into AI agent deployment without proper governance frameworks. Following OpenAI's ChatGPT explosion in late 2022, businesses started building and buying AI agents for everything from software development to ad generation. But many IT departments had no idea what agents were actually running on their networks or what data they could access.
Agent 365 addresses this blind spot by providing administrators with usage analytics - which agents are becoming popular, how many hours of employee time they're saving each week, and crucially, which ones might be overstepping their boundaries. End users get their own analytics dashboard to understand how their AI tools are performing.
