On 1 May 2026, Microsoft Agent 365 became generally available. It is the control platform that lets organisations centrally steer, monitor and secure AI agents. Plus announcements for deeper Intune and Defender integration in June. What does Agent 365 do, who is it for, and how do you start?
On 1 May 2026, Microsoft made Agent 365 generally available. It is a new platform within Microsoft 365 designed for one specific challenge: helping organisations deploy AI agents in a structured, secure and controllable way.
For anyone who has followed Microsoft 365 Copilot in recent years, this is a logical next step. Copilot is the assistant that acts at the request of a user. An agent goes further, autonomously executing tasks, making decisions within a defined scope, and collaborating with other agents to complete a larger process. As organisations deploy more agents, the need for governance arises, and that is what Agent 365 is for.
Agent 365 offers three main capabilities. First, a central overview of all agents active within the organisation, regardless of whether they were developed by your own IT team, supplied by a vendor, or created by an end user in Copilot Studio. Second, policy-based control, determining which agents may use which data, which external systems they may communicate with, and which actions they may perform on behalf of a user. Third, audit and compliance, with full logging of agent actions to Microsoft Purview and Defender.
Agent 365 is available in two ways. First, as part of Microsoft 365 E7, the Frontier Suite covered earlier. E7 bundles M365 E5, Microsoft Copilot, Agent 365 and the Entra Suite for $99 per user per month. Second, as a standalone add-on for organisations that do not want the full E7 bundle.
For most SMB organisations, a standalone Agent 365 licence is financially more realistic than E7. Microsoft has not yet publicly disclosed the exact price for the standalone version, but it is communicated via CSP partners.
Microsoft also announced that in June 2026, a deeper integration will enter public preview between Agent 365, Intune and Defender. Concretely this means context-mapping of agents to specific devices, policy-based runtime controls at the device level, and real-time alerts on suspicious agent behaviour, visible in both the Defender and Intune portals.
For organisations not yet working with agents, the advice is simple. Do not wait for uncontrolled experiments to emerge. Activate Agent 365 in monitoring mode, identify which departments are already experimenting with agents or considering it, and make an initial inventory of which agents you want to allow.
For organisations already actively using Copilot Studio or other agent platforms, the recommendation is to bring all existing agents into Agent 365 and assess whether they meet your new governance baseline. Not every agent built in 2025 will meet the standards you consider logical in 2026.
Want advice on configuring Agent 365 or an evaluation of your current AI strategy? Zarioh helps businesses in the region with the implementation of Microsoft 365 Copilot and Agent 365. Contact us for a no-obligation conversation.
