Microsoft ships Agent 365 GA

Microsoft has not publicly moved Agent 365 into general availability as of April 29, 2026; its own documentation still labels major governance features as preview. (learn.microsoft.com 1) (learn.microsoft.com 2) Agent 365 is Microsoft’s control plane for AI agents, announced at Ignite on November 18, 2025. Microsoft said it would let companies deploy, organize, and govern agents built with Microsoft tools, open-source frameworks, and third-party platforms. (microsoft.com) (news.microsoft.com) In Microsoft’s launch materials, the product centers on five functions: registry, access control, visualization, interoperability, and security. The registry is pitched as a single inventory of agents across an organization, including Microsoft-managed and externally built agents. (microsoft.com) That matters because companies are starting to accumulate large numbers of bots that can read files, call tools, and act on behalf of workers. Microsoft cited an International Data Corporation forecast of 1.3 billion agents by 2028 when it unveiled the product. (microsoft.com) The compliance piece in the user’s prompt is real, but Microsoft’s current public docs do not show a May 1 general-availability launch. A Microsoft Purview page says support for documented Agent 365 protections is “currently rolling out in preview and subject to change.” (learn.microsoft.com) That Purview page says Agent 365 supports auditing, data classification, sensitivity labels, data loss prevention, insider risk management, eDiscovery, data lifecycle management, and Compliance Manager. It also says agent instances are automatically enabled for audit and included in assessments for artificial-intelligence regulations in Compliance Manager. (learn.microsoft.com) Microsoft’s audit documentation also now lists Agent 365-specific events, including AIInvokeAgent, AIExecuteTool, and AIInferenceCall. Those logs are searchable in the unified Microsoft 365 audit log through Microsoft Purview. (learn.microsoft.com 1) (learn.microsoft.com 2) On the security side, Microsoft Defender has a separate AI agent inventory experience, but that page also says the feature is in preview. The inventory covers supported agents from Copilot Studio, Azure AI Foundry, Amazon Bedrock, and Google Cloud Vertex AI. (learn.microsoft.com) Microsoft’s own description of that inventory says admins can query an AIAgentsInfo table to see configuration details and posture signals for Agent 365-managed agents. That is the closest match to the “agent inventory” claim in the prompt, but it is not framed as healthcare-only. (learn.microsoft.com) The healthcare angle appears to come from Microsoft marketing and community guidance aimed at regulated industries, not from a standalone healthcare product launch. A Microsoft healthcare-and-life-sciences blog series on agent governance explicitly targeted regulated sectors such as healthcare and life sciences. (techcommunity.microsoft.com) So the cleaner read is this: Agent 365 is a broader enterprise control layer for AI agents, and Microsoft is still publicly documenting core Purview and Defender governance pieces as preview. If Microsoft plans a May 1, 2026 general-availability step, it has not yet been clearly published in the official sources surfaced here. (microsoft.com) (learn.microsoft.com)