Log in
Agents can now have their own identity, email, OneDrive and Teams accounts, and collaborate just like coworkers.
Microsoft Agent 365 lets you onboard agents, give them the policies and knowledge they need, and let them work in parallel with you to handle tasks like procurement, approvals, research, and updates using the same Microsoft 365 tools you already rely on.
As your use of agents grows, keep full visibility and control. See what they've worked on and understand their impact across your organization as an agent manager.
If you're in IT, you have full visibility and control over access permissions and agent relationships. You can manage all agents from a single unified control plane with the same tools you use now to manage users.
Jeremy Chapman, Microsoft 365 Director, shares how you can adopt autonomous agents at scale across your organization.
Agents that work alongside you.
Assign tasks and get full visibility into what they have worked on using Microsoft 365 tools like Teams and OneDrive. See it here with Microsoft Agent 365.
Automate workflows.
Agents access your data and tools to execute complex tasks. Take a look at Microsoft Agent 365.
Understand agent impact.
Map their actions, connections, and interactions in Microsoft 365 workflows. Get started using Agent 365.
QUICK LINKS:
00:00 — Microsoft Agent 365
01:04 — Agent capabilities
02:48 — Visualize the agent’s impact
03:23 — How it works
04:48 — Agent 365 control plane
07:31 — Zero in on risks
08:18 — Agent map
09:10 — Wrap up
Unfamiliar with Microsoft Mechanics?
As Microsoft’s official video series for IT, you can watch and share valuable content and demos of current and upcoming tech from the people who build it at Microsoft.
- Subscribe to our YouTube: https://www.youtube.com/c/MicrosoftMechanicsSeries
- Talk with other IT Pros, join us on the Microsoft Tech Community: https://techcommunity.microsoft.com/t5/microsoft-mechanics-blog/bg-p/MicrosoftMechanicsBlog
- Watch or listen from anywhere, subscribe to our podcast: https://microsoftmechanics.libsyn.com/podcast
Keep getting this insider knowledge, join us on social:
- Follow us on Twitter: https://twitter.com/MSFTMechanics
- Share knowledge on LinkedIn: https://www.linkedin.com/company/microsoft-mechanics/
- Enjoy us on Instagram: https://www.instagram.com/msftmechanics/
- Loosen up with us on TikTok: https://www.tiktok.com/@msftmechanics
Video Transcript:
-What if an AI agent was truly autonomous, working independently alongside you, with its own email and OneDrive account, capable of joining Teams meetings and conversations to get work done? It means, as a user, you can onboard and manage agents with a unique identity, the right information access, and skills to work on your behalf. These agents can perform the tasks that you define, working autonomously and work with you using the same managed apps and services in Microsoft 365 that you use. And as an IT admin, you have granular control over what agents can do, and knowledge sources they can access. Along with end-to-end visibility into agents in your environment, no matter where they’re created. In fact, with the Agent 365 control plane, we’re extending the same familiar administrative surfaces that you use now to manage people for full visibility, control, and management of agents, while introducing new capabilities.
-So, first, let’s start by looking at what Agent 365 can do from a business user’s perspective. In this case, a coworker has created a procurement agent. And our IT team has approved it, and made it available in our company’s agent store. Now, as a procurement manager, I can find the agent and also set it up with just a couple of clicks. Then once it’s up and running, it contacts me in Teams and asks what I’d like it to do and which tasks to perform. As a procurement agent, it recommends that I give it supplier policies, approved supplier lists, and a procurement playbook. So I’ll do that here with my Teams policy guidelines and just type, use this policy guide for your actions. And then / reference my Zava procurement file.
-Now the agent has what it needs to start working. For interoperability with me, other people, and other agents, it has its own suite of Microsoft 365 apps and a unique account to work on its own. In fact, as an order request comes in from a customer for new laptops, the agent reasons over that request using the instructions I provided. And it can also use contextual business information across Microsoft 365 with Work IQ to find these suppliers, their SLAs, pricing from recent orders, and related documents. Based on the fulfillment time, it even recommends a supplier and asks me if it should proceed. Once I confirm, it creates the purchase order for the laptops and logs that into our purchasing tracker Excel spreadsheet in SharePoint. And right from the comments, like I would at mention any coworker, here I’ve at mentioned the procurement agent for status updates. Agent 365 also makes it easier to visualize the agent’s connections, activities, and impact.
-As a business user, you can see details about the agent, who it’s managed by, its skills, and what it works on in the agent card. You can also see where it fits in the organization, and who it frequently interacts with. Then in the agent activity view, you’ll find its recent sessions with details on actions performed. And clicking into any session activity expands on what was done, the information that was used, and the steps performed to complete its tasks. This is a fully autonomous agent with everything it needs to be effective. In fact, let’s break down the mechanics of how the agent was able to do what it did when it used the Agent 365 control plane.
-The first behind the scenes, once created, the IT approved agent is assigned its own identity in Microsoft Entra and granted access to specific knowledge sources. It’s provided with its own email, calendar, OneDrive, and Teams account, and other services in Microsoft 365. Importantly, it’s also connected to Work IQ, which provides the agent with additional context that’s specific to the jobs it’s performing and the activities by people and other agents around it. But has what it needs to interop with you in the tools that you use every day to get work done.
-Importantly, because it runs on the Agent 365 control plane, it works according to your organization’s security and compliance requirements. For example, least privilege access control ensures that the agent can only access defined content, and nothing more. Also, access can be blocked in real-time based on Conditional Access policies that you have in place. Integrated data security prevents data loss, adhering to your protection policies as it works. And there are also safeguards to keep the agent resilient to targeted attacks. That’s how agents can be onboarded and how they work. Next, as an IT admin, Agent 365 gives you more visibility and control to manage the breadth of agents in your environment, let me show you.
-The Agent 365 control plane in the Microsoft 365 admin center provides an overview of all agents in your organization, with a breakdown by publisher and platform. You can also see whether they were built internally using Copilot Studio, Microsoft Foundry, non-Microsoft platforms, and more. As well as how they’re being used. Below that are recommended top actions to take control, so that you can prioritize your time. Next, to see all of your agents in one place, there’s a complete registry, which pulls in details for security risks, activities, and agent performance into one view. Each agent has comprehensive details. In addition to configuration options, like the data and tools it can access. Information stores it can read from, provisioned compute, graph connectors, tools, and knowledge sources. Then security and compliance provides all of the details for enabled policies with that agent across Microsoft Purview, Microsoft Entra, and Defender.
-Next, in permissions, it goes a step further to display which memberships it has across groups and teams, applications it can access, the SharePoint sites it can use. And detailed permissions across graph API calls. Finally, activity displays information about the agent usage, exceptions and active users. And before agents are available for people to use, as an admin, you’re in full control of validating and approving which agents will appear in your organization’s agent store, here’s how.
-From requests, you can review agents submitted for approval. For example, drilling into this product backlog agent, you can check its configurations, the data it can access, security and compliance protections. And the detailed permissions requested. If everything checks out, you can approve and activate the agent. Then select the right users and groups to access it. In this case, I’ll just keep Mona Kane as the requester. From there, I can apply uniform guardrail policies using customizable templates, like this one, to restrict content sharing. These policy templates leverage Microsoft Entra for access controls, Microsoft Purview to secure data. As well as SharePoint policies, like this one, to enforce specific restrictions on external sharing at the agent level.
-Then I can just review and accept the permissions for the agent, and finally confirm to grant access to its requester. Next, for your running agents, as we saw in the Agent 365 overview, the service automatically and continuously evaluates potential agent risk to alert you of any actions to take. Here, I can zero in on agents with risks. For example, I can see that this comms agent has two risks identified. And when I dig in to see why, it looks like this agent has abnormal sign-in frequency, and was accessed by a user flagged as risky. It’s possible that their account was compromised. And in these cases, Microsoft Entra Conditional Access will automatically block risky agents from accessing resources. And as an admin, you can also block the agent right from here. So it’ll be disabled immediately for current users, and won’t be discoverable for new users.
-Those were single agent operations, but as more agents enter your agent ecosystem with connections to other agents, tools, and knowledge sources, you can see these relationships using the Agent Map. This helps you visually map all agents in your environment across platforms. Importantly, you can see agent connections and multi-agent workflows. Then quickly spot alerts, like this one, for high exception rates. Then drill into view its details, and also take necessary actions. And while today I focused on the experience in the Microsoft 365 admin center, the Agent 365 control plane extends to role-specific views for agents in Microsoft Entra for agent identity and access management, Microsoft Purview for data security protections. And Microsoft Defender for threat detection, investigation, and response.
-And that’s how the new Agent 365 gives you a single control plane to manage agents within the same familiar admin experiences that you’re using today. To get started, from the Microsoft 365 admin center, make sure the Frontier Program is enabled for early access to new AI capabilities. Keep watching Microsoft Mechanics for the latest updates, and thanks for watching.
