Use the prebuilt agent of your choice

With a growing ecosystem of specialized, AI-powered agents in Microsoft 365 Copilot, users now have the flexibility to choose the right tool for the task at hand. Whether you’re preparing for meetings, managing projects, catching up on missed communications, or analyzing data, prebuilt agents are designed to streamline your workflow using your own Microsoft 365 data. This lab offers an open-ended, hands-on opportunity to explore one of these agents, test its capabilities, and apply it to a real productivity challenge from your day-to-day work.

[!NOTE] This course doesn’t include a Microsoft 365 lab tenant with fictitious data. Instead, you must complete this training exercise using your own personal data.

Exercise - Choosing the right agent for your project

This hands-on exercise enables you to take a practical approach to understanding Microsoft 365 Copilot Chat’s prebuilt agents by applying one to a real situation from your own recent work. Think back to a project you were involved with over the past 90 days—something that required coordination, communication, or follow-up. Maybe it was a report you had to write, a team deliverable that stalled, a client request that got buried in email, or a meeting series where decisions were never clearly documented.

Your task is to select a prebuilt Microsoft 365 Copilot Chat agent that you feel can help resolve or gain insights into your project.

[!IMPORTANT] Earlier exercises in this module used the Analyst and Researcher agents. For this exercise, don’t reuse those agents. Instead, use one of the other prebuilt agents to help with your selected project. Doing so ensures you gain experience with the full range of Copilot capabilities and understand which agent is best suited for different types of tasks.

Once you select an agent, you should explore its capabilities using your own Microsoft 365 data, such as emails, meetings, files, or chat threads. Your goal is to see if you can move the project forward or uncover something you missed. At the end of this exercise, you’re also encouraged to try some of the other prebuilt agents to see the types of requests they can help with.

Perform the following steps to see how the prebuilt agent that you chose can function as a real productivity partner by solving a problem that matters to you:

  1. In Microsoft Edge, open a new tab and enter the following URL: https://m365.cloud.microsoft.com

  2. In Microsoft 365 Copilot, in the navigation pane, select All agents, and then in the Agent Store window, in the Built by Microsoft section, select the agent of your choice. Remember, don’t use the Researcher or Analyst agents. Try a different agent for this exercise.

  3. The goal of this exercise is to interact with the agent using your own data. You should begin by reviewing the agent’s built-in prompt suggestions. Try one or two of the sample prompts that appear on the agent page.

  4. After testing the sample prompts, run three to five queries or actions the agent supports. Explore what data the agent can access, such as your calendar, emails, files, or organizational data. Make sure the prompts you enter align with your project and the agent you chose. The following sample prompts can help spur ideas:
    • Analyze a team activity (if the agent supports it)
    • Identify key files, tasks, or trends
    • Ask how to onboard a new system
    • Help brainstorm an upcoming project
    • Provide advice on how to progress through your career path
    • Provide feedback on a writing project
  5. Now ask your agent to create a concrete output, such as a summary, list, document, or action plan. For example:
    • Review missed messages and generate a follow-up plan
    • Draft a status update or meeting preparation document
    • Create a report that summarizes email threads or conversations
  6. Take this opportunity to try out some of the other prebuilt agents. Start by reviewing the sample prompts they provide, which are indicative of the types of requests they specialize in. Run some of the sample prompts or enter some of your own custom prompts to get a feel for the types of requests that some of the other agents can help with.