Something interesting happened quietly on May 1st. Microsoft made Agent 365 generally available, and with it came an official acknowledgment from one of the world’s largest enterprise software companies: businesses are running AI agents they cannot account for, and that needs to change.

Agent 365 is Microsoft’s governance and security control plane for AI agents. The pitch is simple. IT and security teams need to see every agent operating inside their organization, regardless of which cloud it runs on or who deployed it. Without that visibility, you have what Microsoft is now calling “shadow AI” — agents running on employee devices, accessing company data, taking actions on behalf of users, and doing all of it completely outside any oversight.

The Shadow AI Problem Nobody Wanted to Admit

The timing of this launch tells you something. For the past two years, the conversation in enterprise AI has been dominated by enthusiasm: what agents can do, how fast they can work, how much headcount they can replace. What got less attention was the governance question. Who approved these agents? What data can they access? What happens when one goes wrong?

Agent 365 addresses that gap by giving IT teams a live dashboard of every registered agent in their environment. It surfaces key metrics like total agents deployed, active users, runtime hours, and emerging risk signals. It also syncs with AWS Bedrock and Google Cloud, so the platform works even in mixed-cloud environments — which is almost every large enterprise at this point.

The shadow AI detection capability is particularly striking. Microsoft has built detection for unmanaged agents running on Windows devices, starting with OpenClaw and Claude Code, with GitHub Copilot CLI expanding support coming. If someone on your team has deployed a local AI agent that nobody approved, Agent 365 will find it.

Pricing and Packaging

Agent 365 is available standalone at USD $15 per user per month. It is also included in Microsoft 365 E7, the new “Frontier Suite” that bundles Microsoft 365 Copilot, Microsoft 365 E5, Microsoft Entra Suite, and Agent 365 for $99 per user per month.

Microsoft has also launched Windows 365 for Agents in public preview for US customers, which provides policy-controlled compute environments for running agent workloads within defined guardrails.

Partners including Accenture, KPMG, Adobe, Zendesk, and n8n are already integrated with the platform to support enterprise deployments at scale.

What This Means for Business

The launch of Agent 365 marks a shift in how the industry is thinking about AI agents. They are no longer experiments. They are infrastructure. And infrastructure needs governance.

A few things business leaders should take from this:

The agent inventory problem is real. Most businesses deploying AI tools today have no complete picture of what is running, what it can access, or who authorized it. Shadow AI is not a future risk. It is a present one, and Microsoft building a product to solve it confirms that at scale.

Governance is now a procurement question. When evaluating AI vendors or deciding how to structure your AI operations, asking “how do we govern this?” is no longer optional. You need to know who can deploy agents, what data they touch, and how you would shut one down if something went wrong.

Multi-cloud governance is the default. Agent 365 supports AWS and Google Cloud out of the box. That tells you where enterprise AI is actually running: everywhere, not just in one vendor’s ecosystem. Any governance strategy that only covers Microsoft tools is already incomplete.

The compliance clock is ticking. Regulatory frameworks for AI are developing fast across the US, EU, and Australia. Businesses that have no agent inventory today will struggle to demonstrate compliance when those frameworks require it.

For businesses already working with Enterprise DNA on AI deployment through Omni, governance has always been part of the conversation. The question of who controls the agents, what they can access, and how performance is measured sits at the foundation of any responsible AI implementation.

The arrival of a Microsoft-backed governance product at this price point signals that the market is moving from “should we deploy AI agents?” to “how do we run them responsibly?” That is a healthier question for everyone.

If your business is still working out where to start with AI governance, Omni Advisory offers fractional AI advisor support to help you map what you have, what you need, and how to deploy it without the blind spots.