[Digital Today reporter Chi-gyu Hwang] Microsoft said on Thursday it has officially launched Agent 365, a platform for integrated management of AI agents within work environments.

Agent 365 is an integrated control platform that detects, manages and protects AI agents across an organisation from a single place.

Agents running in Microsoft Copilot, Teams and Microsoft 365, as well as agents from external partners, are included in its management scope.

The company said the spread of AI agents is emerging as a new security threat. As more employees install agents such as OpenClaw and Claude Code directly on devices or adopt external agents provided as software as a service, these agents can create a “Shadow AI” problem by modifying code or accessing sensitive information outside IT team oversight.

Microsoft added detection features for local and cloud agents using Microsoft Defender and Intune. It currently supports detection of the OpenClaw agent on Windows devices, and will expand coverage to GitHub Copilot CLI and Claude Code.

From June, Agent 365 will also provide an asset context mapping feature for each agent. This will allow organisations to see at a glance the devices on which agents run, connected MCP servers, related accounts and accessible cloud resources, the company said.

It also added support for multi-cloud environments. Integration with AWS Bedrock and Google Cloud will allow IT teams to automatically detect and list agents deployed across multiple cloud platforms.

Microsoft will soon support lifecycle management functions such as starting, stopping and deleting agents. Support for ecosystem partner agents has also expanded. Agents from partners including Genspark, Zensai, Egnyte and Zendesk, as well as agents based on Kasisto, Kore and n8n, can be managed in Agent 365 without separate integration work.

Microsoft also introduced “Windows 365 for Agents”, a cloud PC environment dedicated to AI agents, in a public preview in the United States. Windows 365 for Agents is designed so agents run in a policy-controlled environment managed by Intune, Microsoft’s endpoint security product. Network security features have also been strengthened. It can inspect agent traffic at the network layer to block access to unapproved AI services and prevent malicious prompt attacks.