Microsoft launches Agent 365 to oversee AI agents
Mon, 4th May 2026 (Today)
Microsoft has made Agent 365 generally available. The product is designed to help organisations oversee AI agents across business systems.
The launch comes as companies deploy growing numbers of AI agents through workplace software, developer tools and cloud platforms, creating new oversight and security challenges for IT and risk teams.
Agent 365 is positioned as a single management layer for agents that act on behalf of users and those that operate with their own credentials. It is intended to let organisations observe, govern and secure agents across local devices, software services and cloud environments using existing administration and security workflows.
Microsoft also outlined several new preview features and integrations tied to the service. These include discovery tools for local and cloud-hosted agents, support for a broader range of software-as-a-service agents, and integration with Windows 365 for Agents, a managed cloud PC environment for agent-based workloads.
Wider visibility
A central concern for customers is the growth of so-called shadow AI, in which agents are installed or used outside formal controls. Users are already running local agents such as OpenClaw and Claude Code on devices, while cloud-based agents are being built on platforms including Microsoft Foundry, AWS Bedrock and Google Gemini Enterprise Agent Platform.
To address that, Microsoft is adding discovery and management functions through Microsoft Defender and Intune. Organisations will be able to identify local AI agents on Windows devices, beginning with OpenClaw and later extending to tools including GitHub Copilot CLI and Claude Code.
Customers in Microsoft's Frontier programme can already see whether OpenClaw agents are in use, which devices they run on, and apply Intune policies to block common methods of operation. A registry of local agents will also appear in Defender and Intune, giving IT and security teams a shared inventory.
Further additions are planned for public preview in mid-2026, including asset context mapping that links each agent to the device it runs on, configured MCP servers, associated identities and reachable cloud resources. This is intended to help security teams assess exposure, investigate file access and network behaviour, and define custom detections.
Cross-platform reach
Beyond Windows endpoints, Agent 365 now includes public preview support for registry synchronisation with AWS Bedrock and Google Cloud connections. This will allow IT teams to discover and inventory cloud agents across those platforms, with basic lifecycle actions such as starting, stopping and deleting agents expected to follow.
Microsoft is also extending Agent 365 to a broader software ecosystem. The service works with prebuilt agents in Microsoft 365 Copilot and Teams, agents created with Copilot Studio or Foundry, and partner-built agents.
Named partner integrations include Genspark, Zensai, Egnyte and Zendesk, as well as agent factory providers such as Kasisto, Kore and n8n. These agents can be managed through Agent 365 without additional integration work by customer IT or security teams.
Network controls
AI agents can move far faster than human users and may connect to risky web destinations or unsanctioned AI services unless controls are in place. Agent 365 now extends Microsoft Entra network controls to Microsoft Copilot Studio agents and to agents running on user endpoint devices, including local agents such as OpenClaw.
According to Microsoft, these controls are available now as part of the general release. They are intended to help security teams inspect traffic at the network layer, restrict outbound connections to approved destinations, filter risky file movement and block prompt-based attacks before they trigger harmful actions.
Managed environment
Microsoft also linked the general release of Agent 365 to Windows 365 for Agents, which is in public preview in the United States. The service introduces a version of Cloud PCs intended for agent workloads and managed in Intune.
With Agent 365, organisations can see which agents are connected to those cloud systems through the Microsoft 365 admin centre. The combination is meant to help customers move from basic visibility and governance to running agents in managed production settings.