Microsoft is ramping up its investment in enterprise AI agent workflows with the launch of a new centralized AI agent command center: Agent 365.

Announced today at Microsoft Ignite 2025, Agent 365 offers unified visibility and governance for AI agents deployed across enterprises. The platform consolidates telemetry, dashboards, and alerts into a single interface for streamlined oversight.

“As the number and complexity of agents grow, organizations face a new challenge: how to responsibly manage and govern them at scale without rebuilding the trusted systems they rely on,” said Charles Lamanna, President of Business Applications and Agents at Microsoft.

He added that managing agents in the enterprise should mirror how people are managed—through structured oversight and clear accountability.

To address this, Agent 365 integrates five foundational pillars: registration, access control, observability, interoperability, and security.

Built on Microsoft Entra—a comprehensive identity and access management suite—IT teams gain a single source of truth for every agent within the organization. This includes a unified registry tracking all agents and their associated roles across IT, development, security, and business operations.

Every agent in the system must have a unique agent ID, enabling access management and restrictions similar to user role assignments. IT can predefine safeguards for creating, onboarding, and managing agents. Policies within Entra enforce adaptive, risk-based roles that respond dynamically to context and threat levels, helping detect and contain rogue agents in real time.

On the front end, users gain access to an advanced analytics dashboard that maps relationships among agents, users, and resources. Engineers and IT teams can drill into custom metrics to understand how complex networks of software and personnel interact. The dashboard surfaces agent performance indicators—such as speed, quality, and reliability—alongside detailed logs, reports, and discovery tools to support security and ethical audits of agent behavior.

Microsoft emphasizes that the new command center is vendor-agnostic, designed to work seamlessly with agents built on Microsoft platforms, open-source frameworks, or third-party systems—whether hosted on Azure or partner clouds.

The solution is tightly integrated with Microsoft Defender, the company’s cybersecurity suite, and Purview, its unified data governance and compliance platform, delivering end-to-end visibility and control over both agents and data.

Purview mitigates potential AI-related data exposure risks by enforcing policy controls that prevent agents from processing sensitive information. Defender enables organizations to detect and block known vulnerabilities and emerging threats targeting agents, while Entra provides real-time protection against active attacks.

Governing and Securing Copilot

As Copilot adoption expands, organizations increasingly require tools to monitor and secure its usage. In response, Microsoft has released Baseline Security Mode, now generally available.

This feature helps organizations protect their Microsoft 365 environments using recommended configurations. After an onboarding experience, IT and security teams can identify policy gaps, simulate changes, and deploy protections across common Microsoft applications—including Office, SharePoint, Teams, and Entra.

Additionally, Microsoft introduced a new Agent Dashboard, currently in preview, offering centralized reporting on agent adoption across the organization. IT leaders and executives can use it to analyze trends, assess group-level uptake, and track individual agent actions—enabling data-driven decisions around governance and impact.

Purview has also been updated with Data Loss Prevention (DLP) for Microsoft 365 Copilot, now in preview, to safeguard sensitive information. If a user prompt contains confidential data—such as credit card numbers or personally identifiable information—the new Prompt Protection feature will block Copilot and connected agents from responding, preventing such data from being used in model training or web searches.

Also in preview, Purview’s Data Security Posture Management (DSPM) now helps IT administrators identify and remediate overly permissive sharing links, reducing the risk of unintended data exposure. Meanwhile, Purview AI Observability—integrated into DSPM and also in preview—enables IT teams to monitor all agent activities and empowers security teams to make informed decisions to mitigate risks.

New Agents for Business Users and IT Admins

Beyond the Agent 365 command center, Microsoft unveiled several new AI agents designed to help employees extend their expertise and productivity.

Powered by Work IQ—the intelligence layer behind Microsoft 365 Copilot, Microsoft’s AI-powered assistant embedded across its productivity suite—these new agents aim to simplify workforce management, support skill development, and strengthen employee connections.

Among them is a Workforce Insights Agent that delivers real-time visibility into workforce composition, tenure, and location, empowering leadership with data-driven insights for team planning. A People Agent helps users identify the right colleagues to contact based on past interactions, fostering internal collaboration. Meanwhile, a Learning Agent offers personalized micro-lessons, tips, and curated courses to enhance employee skills and AI literacy.

Sales teams will benefit from a new Sales Development Agent, now in preview. Acting as a fully autonomous sales assistant, it conducts research, qualifies leads, and engages prospects outside business hours—escalating complex issues to human reps when needed.

In Teams—the collaboration hub combining chat, video meetings, and file sharing—a new Teams Admin Agent (in preview) automates repetitive administrative tasks. It can securely perform functions like monitoring meetings or managing user accounts directly within Teams.

Likewise, a new SharePoint Admin Agent, also in preview, equips IT administrators with AI-driven insights and automation. According to Microsoft, it helps identify inactive or orphaned sites, track over-shared content and permission sprawl, and automatically apply policies to archive or restrict access as needed. Admin teams can also leverage it to better understand the growing footprint of AI agents as Copilot adoption accelerates.