Table of Contents
Planner Agent Helps People Understand How to Get Tasks Done
Message center notification MC1250279 (12 March 2026, Microsoft 365 roadmap item 511820) announces that the Project Manager agent, part of Microsoft Planner, will be renamed to be the Planner agent. More importantly, the agent will be available to users with Microsoft 365 Copilot licenses who have either a Planner premium or Planner basic plan. Previously, the agent was confined to users with Planner premium licenses. Apparently, the move is all part of the grand plan to align naming and expand availability of Copilot services.
Targeted release tenants will see the change before the end of March 2026. My tenant is targeted and I see the agent show up in plan rosters. General availability is due to be completed by early May 2026.
How Does the Planner Agent Work?
The news about an agent renaming is exciting enough for one day. After calming down, I looked for the agent documentation and learned that the Planner agent is supposed to get work done faster and can “create your plan based on goals, execute tasks, and act on feedback.” Essentially, the Planner agent takes the details of a task and uses that information to generate an AI outline describing how to accomplish the task.
To invoke the agent, you create a task as normal and then assign the Planner agent to the task (Figure 1). Planner automatically adds the user that assigns the agent to “help review the response of the AI agent.”
Planner then queues the task for AI processing. After a couple of minutes, the task is picked up by a background task and processed. Planner uses a special label to indicate the state of processing. As you can see in Figure 1, a task starts out as Queued, then progresses to In Progress, and finally ends up as Ready. Task Assignees receive email to notify them when the Planner agent completes its processing.
Any of the plan members can see the advice generated by the Planner agent through the AI agent activity tab. Currently, the tab is not visible in the Planner mobile client, so you must use the browser client or Teams app to see the information. Figure 2 shows the Teams app being used to view the recommendations generated by the Planner agent.
Refining AI Recommendations
The recommendations generated by the Planner agent are in a Loop component. Given that the information provided to the agent in a task might not be complete, it’s likely that some further interaction is necessary to refine the recommendations to a point where they are valid, useful, and actionable.
Plan members can edit the Loop component through the task or using the Loop browser app. Figure 3 shows the recommendations for the task being edited in the Loop app. The advantage of using Loop to hold the agent recommendations is that Loop supports simultaneous edits to the information with near-instantaneous updates.
In this case, the input comes from a plan member who is not assigned to the task. Remember, all plan members have equal access to the tasks in a plan. If you want to keep something confidential, put it in a separate plan.
Regenerating Planner Agent Recommendations
At any point in the editing process, you can regenerate the AI recommendations. The same queueing process occurs to submit the updated information (from notes added for the task and the text in the Loop component) for processing. The Loop component is then refreshed with the updated recommendations. This sequence continues for as many iterations as necessary. Task chat (Figure 4) is usually helpful to resolve issues raised in the AI-generated text.
Hopefully, the final outcome is a helpful outline explaining how to achieve the task. The recommendations in the Loop component can be copied into a Word document, shared with other users, or printed as a PDF.
Precise Framing of Tasks is Critical
As always, the quality of the input determines the quality of the output. If a task is described in loose terms with no clear direction about what’s expected and no direction about restraints that need to be respected, the AI output will be bland, generic, and unhelpful. Precise framing of the task with some detailed notes about how to approach the work will give the Planner agent better guidance about what’s expected and the output shouldn’t be AI slop. Treat the agent output like you’d handle the work of a junior associate and be prepared to iterate to work to improve its output, and the Planner agent might just work for you.
Learn about managing Planner and the rest of the Microsoft 365 ecosystem by subscribing to the Office 365 for IT Pros eBook. Use our experience to understand what’s important and how best to protect your tenant.