Microsoft on Tuesday (Nov. 18) introduced Microsoft Agent 365, a new system aimed to help companies track and control the growing number of AI agents used in everyday work.
The company said organizations are starting to rely on agents to handle tasks across support, operations and productivity tools, and that they need a single place to manage them. Microsoft cited IDC research estimating that more than one billion agents could be active inside businesses by 2028.
Agent 365 gives companies a unified view of all the agents running in their environments. It includes a central list where IT teams can see approved agents as well as “shadow” agents that may have been created without permission. Companies can set rules about what each agent is allowed to access, using Microsoft Entra to manage those permissions.
The system also provides a dashboard that shows how agents are being used, who they interact with and which tools they connect to. The platform supports agents built with Microsoft tools as well as third-party frameworks, and security features from Defender and Purview extend threat protection and audit controls to all registered agents.
The launch signals that many companies are moving past small experiments and now need ways to manage larger groups of agents. Microsoft is positioning Agent 365 to bring structure and accountability to these deployments so businesses can scale them safely and measure their impact.
The visibility and reporting tools could also help companies determine where agents are working well and where adjustments are needed. As agent adoption grows, Microsoft said organizations are increasingly worried about unauthorized agents operating without oversight, which Agent 365 aims to address.
The system builds on Microsoft’s earlier step from May, when it introduced the Agent Store. That marketplace let users browse and install prebuilt agents designed for everyday tasks in Microsoft 365 apps. PYMNTS reported that the store launched with more than 70 agents and was aimed at helping workers bring AI into their existing workflows without learning new tools.
Source: https://www.pymnts.com/