Copilot Studio: Meetings → Planner Tasks

Key insights

• Copilot Studio + Granola MCP: The video shows how a Copilot Studio agent reads meeting notes from a Granola MCP server and turns action items into Microsoft Planner tasks, automating follow-ups.
  
• Granola MCP: Granola captures and summarizes meetings from Teams, Zoom, and Google Meet, extracts action items and owners, and exposes that data via an MCP server for agents to query.
  
• MCP (Model Context Protocol): MCP is the standard interface agents use to call external tools. In Copilot Studio you add an MCP server as a tool, set authentication, and test it so the agent can query meeting data directly.
  
• Planner integration: The agent can add Planner actions like "Create a task", map tasks to a plan and bucket, set the owner and due date (for example using the meeting deadline), and populate priority and titles automatically.
  
• Setup steps: Create a Copilot Studio agent, add the Granola MCP tool, turn off broad web search to limit results to meeting data, ask for this week’s action items, add the Planner task action, configure plan and bucket, then test the agent.
  
• Benefits & best practices: This approach reduces manual task entry, keeps follow-ups tracked in Microsoft 365, and saves time. Approve Granola and Planner actions once, use meeting deadlines as due dates, use clear naming (owner’s first name + task), and verify required Microsoft 365 Copilot/Planner licenses are in place.
  

Video at a glance

The YouTube tutorial "Copilot Studio + Granola MCP: Turn Meetings into Planner Tasks" by Anders Jensen [MVP] demonstrates how to automate the conversion of meeting action items into Microsoft Planner tasks. In clear steps, the creator builds an agent in Copilot Studio, connects a Granola MCP server as a tool, and configures the agent to create tasks with owner and due date in Microsoft Planner. The video highlights practical settings such as turning off "Search all websites" so the agent uses meeting notes only, and it shows the approval process the first time the agent runs. Overall, the clip is a concise walk-through aimed at people who want to reduce manual follow-up work after meetings.

First impressions emphasize that this workflow stitches together three layers: meeting intelligence, agent orchestration, and task execution. Anders demonstrates how meetings captured across platforms like Teams, Zoom, and Google Meet can be summarized by Granola and exposed to an agent via the MCP protocol. Then the agent maps extracted action items into Planner tasks, including assigning an owner and using meeting deadlines as due dates. The result promises to save time for users who manage many recurring meeting follow-ups.

How the solution works

Anders shows that Granola MCP acts as the meeting-intelligence layer by summarizing transcripts and extracting action items across multiple meeting platforms. Next, a Copilot Studio agent is created and the Granola MCP server is added as a tool in the agent’s Tools area, where you provide the server URL and authentication. The agent then queries the server for action items, and the video demonstrates adding the Planner action "Create a task" and pointing it to a chosen plan and bucket. Finally, a small set of instructions makes the agent format task titles with the owner’s first name and populate due dates from meeting deadlines.

The tutorial also highlights the first-run approval model in Copilot Studio: when the agent uses Granola and Planner actions for the first time, the user must explicitly approve those connectors. Once approved, a single prompt can convert a week of meeting notes into structured tasks with priority and deadlines automatically filled in. Anders notes that this behavior reduces repetitive clicks while maintaining a level of manual control during initial setup. In addition, he teases a follow-up demo that explores the Granola API for more granular control.

Step-by-step setup demonstrated

In the recorded chapters, Anders starts by creating the agent and then shows the precise flow to add the Granola MCP server as a tool inside Copilot Studio. He recommends disabling broad web search to constrain answers to meeting content only, which improves relevance and data governance. After querying for this week’s action items, Anders builds in the Planner action and configures mapping to a plan and the "To do" bucket so tasks land where teams expect them. Throughout, he tests the agent and validates that task titles, owners, and due dates follow the instructed format.

The demo covers practical touches like naming conventions for task titles and the importance of mapping owner fields correctly to avoid misassignment. It also shows that users can run the agent iteratively and refine prompts to improve task extraction. The setup is repeatable for teams who want to standardize how meeting outcomes become tracked work. Moreover, Anders emphasizes the value of preview testing in Copilot Studio before enabling full automation in a production environment.

Tradeoffs and challenges

While the automation saves time, the video also implicitly points to tradeoffs around trust, accuracy, and control. For example, automated extraction can misinterpret informal language or implicit ownership, so teams will likely need a review step to avoid assigning tasks incorrectly. In addition, security and compliance considerations matter because giving an agent access to meeting transcripts and Planner raises permission and data residency questions. Anders counters some risk by showing how to limit the agent’s scope with the "Search all websites" setting and by requiring explicit approvals for each action on first use.

Technical and operational challenges also appear in the tutorial: mapping complex meeting outcomes into discrete Planner tasks can be ambiguous, and deadlines derived from conversations are sometimes vague. There is also the practical need for proper authentication and role-based access in Microsoft 365, which can complicate deployments for larger organizations. Finally, maintaining synchronization between the Granola backend and agent tools may require attention as connectors evolve, so teams should plan for periodic testing and updates.

Implications and next steps

For organizations that run many meetings and struggle with follow-up, the pattern Anders demonstrates — Granola MCP into Copilot Studio into Microsoft Planner — offers a clear path to reduce manual effort. It makes meeting outcomes actionable and traceable, and it can be adapted with additional rules or API-level controls for more precise workflows. However, teams should weigh the benefits against necessary governance steps, testing, and ongoing monitoring to ensure tasks remain accurate and appropriately assigned.

In closing, the video is a practical how-to that balances usability with caution. Developers and admins interested in this approach will find a repeatable setup and a sensible starting point, while recognizing that more advanced use cases will require deeper integration work and governance. Anders Jensen’s tutorial thus serves as both a demonstration and a prompt to evaluate how automation can be introduced thoughtfully into meeting-driven workstreams.

Keywords

Copilot Studio Granola MCP, Turn meetings into Planner tasks, Microsoft Copilot meetings to Planner, Teams meetings to Planner automation, Granola MCP integration with Planner, Automate meeting task creation Planner, Copilot Studio tutorial Planner workflow, Microsoft Planner meeting task automation