CrowdHum · MCP server

Use CrowdHum from your favorite AI

CrowdHum is now available as an MCP server. Connect it to the AI assistant you already use and let it run your polling for you. Point it at the PowerPoint for your next talk and ask for a poll set. After the room votes, ask it to analyze the results. No copy-paste, no tab-switching.

Before your talk

Turn a deck into polls

Point your assistant at a PowerPoint or an agenda and ask for polls. It reads your session and proposes ready-to-run questions you can launch in one click.

After your talk

Analyze your results

Ask your assistant to break down how the room voted: the themes, the splits, and the one takeaway worth repeating on stage.

Before you start

Connect it to your AI of choice

CrowdHum is listed in the ChatGPT plugin directory, so there it is a two-click add. It also works with any other AI assistant that supports MCP. Pick the one you use, or follow the same pattern in your own client.

Once you're connected, try these

Talk to your assistant in plain language. A few prompts to start:

What CrowdHum can do

The MCP server gives your assistant ten tools. It reads and writes only your own CrowdHum polls and sessions.

Read

  • List visualization modes: the poll styles (word cloud, bubbles, 2×2, Q&A…).
  • List my sessions: find a session by name and get its room code.
  • List my polls: your polls, newest first.
  • Get a session's details: the polls in a room and their vote counts.
  • Get room links: the voter and presenter links for a room.
  • Get a poll's results: live or final tallies, broken down by option.
  • Get a session's AI recap: themes, splits, and the takeaway.

Create & manage

  • Create a poll: a new poll in a fresh room, with a link to share.
  • Add a poll to a session: one QR, many polls in the same room.
  • Invite a co-owner: let another Pro user co-edit and co-present.
Using a different MCP client? Any client that speaks the Streamable HTTP transport can connect. Point it at https://mcp.crowdhum.com. If your client doesn't support OAuth, generate a personal token at crowdhum.com/connect and send it as an Authorization: Bearer <token> header instead.