Recall.ai

@recall.ai list all active bots and show me which meetings they're currently in

Based on your Recall.ai account, there are currently 3 active bots: Bot #4821 is in the 'Q1 Planning Review' Zoom meeting (started 14 minutes ago), Bot #4819 is in 'Customer Success Sync' on Google Meet (started 2 hours ago), and Bot #4823 just joined 'Engineering Standup' on Microsoft Teams. All three are recording. Bot #4819 appears to have been running longer than typical — you may want to check if that meeting ended.

This example shows the MCP's ability to surface bot status across your organization's meetings. Useful for auditing which calls are being captured, but note that bot metadata doesn't include participant names or meeting content — you'll need to retrieve chat messages or recordings separately for that context.

@recall.ai create a bot for the Zoom meeting at zoom.us/j/123456789 and start recording immediately

Bot created successfully (ID: #4827) and deployed to the Zoom meeting. The bot has joined as 'Recall Bot' and recording started at 10:42 AM UTC. It will capture audio, video, and transcription. The bot is now visible to other participants in the meeting. You can retrieve chat messages or stop the recording anytime by referencing bot ID #4827.

This demonstrates the core workflow: programmatically injecting a bot into a live meeting. Important: the bot appears as a named participant, so meeting attendees will see it join. You need the meeting URL and appropriate Recall.ai API quota. Recording starts immediately unless you specify otherwise — plan for storage costs on longer calls.

@recall.ai pull the chat messages from bot #4821 and summarize any action items or decisions mentioned

Retrieved 47 chat messages from bot #4821 in the 'Q1 Planning Review' meeting. Key action items identified: Sarah to finalize the budget deck by Friday, engineering team committed to shipping the API v2 beta by March 15, and the group agreed to move the launch date to April 2 pending legal review. Two participants shared links to Notion docs for follow-up. No explicit decisions were recorded in chat about the pricing model — that discussion likely happened over audio.

This example pairs the MCP's chat retrieval tool with AI reasoning to extract structured insights. The MCP only returns raw chat messages (text, sender, timestamp) — the summarization happens via the AI. Useful for post-meeting synthesis, but remember chat is just one data stream; critical decisions may occur in voice/video only.

When Recall.ai makes sense for distributed sales QA

A 6-person sales team running 40+ Zoom demos a week needs call recordings for coaching and CRM notes. Recall.ai is the right call here because you can spin up bots programmatically, pull transcripts into your workspace, and delete them when the deal closes. The API_KEY auth means your ops lead sets it once and the whole team shares access through Switchy. The chat-message tool is useful if your reps paste links or pricing in-call. Trade-off: if you're only recording internal standups or retrospectives, a native Zoom recording is simpler. But for client-facing calls where you need post-call automation (summary to Slack, transcript to HubSpot), this MCP pays off. Worth it if you're recording more than 10 external calls a week.

Using Recall bots for onboarding knowledge capture

A 3-person CS team onboards new enterprise customers with live walkthroughs, and they want a searchable library of common questions and setup patterns. Recall.ai works well because you can create a bot per onboarding call, retrieve the transcript and chat messages afterward, and feed them into a shared knowledge base in Switchy. The remove-bot and delete-bot tools let you clean up after each session so you're not paying for idle bots. The threshold: if your onboarding is mostly async (Loom videos, help docs), this is overkill. But if you're doing 5+ live sessions a week and manually taking notes, the bot handles capture while your team stays present. Best fit for teams who want call data in their workspace, not locked in a vendor dashboard.

When Recall.ai is too much for internal meetings

A 12-person startup runs a weekly all-hands and wants recordings for remote teammates in other time zones. Recall.ai can do this, but it's probably not the best tool. You'd create a bot, start recording, stop recording, then retrieve the transcript—but you're adding API complexity for a use case that Google Meet or Zoom handles natively with one click. The chat-messages tool is useful if people drop links or action items in chat, but most teams already have that in their meeting platform. The break-even point: if you need to pipe meeting data into another system (a wiki, a project tracker, a custom dashboard), then the MCP makes sense. Otherwise, stick with your video platform's built-in recording and save the API_KEY for a workflow that actually needs programmatic control.