How to brainstorm with AI agents in a meeting

Give the humans a question first

Start a Rooms meeting around one open question: which approach should we try, what is missing from a proposal, or what evidence would change our view? Invite people who can judge the result and one agent with a bounded role. Avoid asking several agents to solve the whole meeting at once.

Separate idea generation from checking

Let participants sketch their ideas before asking an agent to organize them. Then give it a specific job: ‘Group these approaches without dropping minority views. Mark disagreements and questions we have not answered.’ Ask for a short response first and a board when the group needs a common view.

Use agents to investigate a gap

When the discussion reaches an unknown, ask an agent with the appropriate tools to investigate it. Request sources and distinguish retrieved evidence from inference. Web access and other external tools depend on the selected teammate and its environment. If no source is available, keep the claim unresolved instead of asking for a confident guess.

Close with an explicit decision

At the end, ask for a draft of what the team agreed, what remains open, and who owns the next step. Have a participant confirm those details. This is a suggested facilitation workflow for Rooms, not evidence that the product automatically makes decisions, prevents interruptions, or guarantees shorter meetings.

Explore more

• An agent-native meeting platform

  Rooms brings people and AI agents into the same browser meeting, with conversation context, attributed replies, and shared boards.

• How to bring your AI agent into a meeting

  Invite an outside agent to a Rooms meeting, connect it using the supplied instructions, and verify that it can participate.

• Bring Claude Code into your Rooms meeting

  Connect Claude Code to a Rooms call so your team can ask questions and review its explanations and boards together.