A Control Plane for the AI Agent Era
If you've been watching the enterprise AI space, you know the chaos that's been quietly building. Companies are deploying AI agents from a dozen different vendors, on different clouds, built on different platforms — and nobody has a real handle on what those agents are actually doing. Microsoft's answer to that mess is Agent 365, which became generally available for commercial customers today.
Think of it as a control room. One centralized place where IT, security, and business teams can see, manage, and govern every AI agent running inside an organization — whether it was built on Microsoft's own tools or someone else's entirely.
And honestly? The timing feels deliberate.
What Agent 365 Actually Does
The product is built around three ideas: observability, governance, and security. Microsoft's framing here is pretty blunt — "you can't govern what you can't see." That's the whole premise.
Observability: Finally Knowing What's Out There
Microsoft Defender now includes a dedicated AI agent inventory. It discovers and tracks agents built on Microsoft Copilot Studio, Microsoft Foundry, AWS Bedrock, and GCP Vertex AI. So it's not just a Microsoft-ecosystem play — it's designed to surface agents running across your entire stack, regardless of where they came from.
Governance: Identity for Every Agent
Every agent registered in Agent 365 gets a Microsoft Entra Agent ID. That's a real identity — one that enables conditional access policies, lifecycle management, and all the control mechanisms enterprises already use for human users. It's essentially bringing the same identity framework you'd apply to an employee and extending it to your AI agents.
Security: Stopping Data Leaks Before They Happen
Microsoft Purview adds inline data loss prevention that inspects prompts before they're processed. Not after. That distinction matters a lot when you're dealing with sensitive enterprise data.
Pricing and Packaging
Agent 365 runs $15 per user per month as a standalone add-on. But Microsoft is also bundling it into a new Microsoft 365 E7 tier at $99 per user per month — which combines Microsoft 365 E5, Copilot, Agent 365, and the Entra Suite. That bundle represents roughly a 15% discount compared to buying everything separately.
The Partner Ecosystem Is Already Moving
The launch didn't come alone. Genspark announced a strategic partnership this week, embedding its AI agents for presentations, spreadsheets, and documents directly into Microsoft 365 and Agent 365. Kore.ai was named a launch partner back at Microsoft Ignite in November 2025. Open-source workflow platform n8n is now a first-class Agent 365 citizen — agents built on n8n get governed Microsoft 365 identities, which is a big deal for teams already using it. Reply also came on as a launch partner in April.
And the multicloud story is real, not just marketing. The documentation covers building and deploying agents on AWS Elastic Beanstalk and Google Cloud Run while still using Microsoft Entra and Graph for identity and governance.
The EU AI Act Is Four Months Away
Here's a detail that probably explains some of the urgency around this launch, especially for European enterprises. The EU AI Act becomes fully applicable on August 2, 2026 — just three months from now. Agent 365's registry, identity controls, and compliance integrations could serve as a foundation for meeting some of the Act's requirements around risk management, transparency, and human oversight. Though organizations will need to actively do the work of mapping those capabilities to specific regulatory requirements — it's not an automatic compliance solution.
Windows 365 for Agents, announced in January, extends the platform further into cloud PC environments, adding check-in/check-out models for short-lived agent workloads and human-in-the-loop controls for sensitive operations.