How to Connect Recall.ai to Claude (and What It Can't Do)
Recall.ai is a different animal from the other tools people try to plug into Claude. It isn’t a notetaker you install — it’s the meeting-bot API that notetakers are built on: you create bots programmatically, they join Zoom, Meet, Teams, or Webex calls, and they stream back recordings, transcripts, and metadata. So “connecting Recall.ai to Claude” usually means one of two people is asking: a developer building on the API, or a team lead hoping to chat with their meeting archive. The answer differs for each.
The official MCP server — built for the developer
Recall.ai runs a first-party, hosted MCP server, documented here, with regional endpoints such as:
https://us-east-1.recall.ai/mcp(us-west-2, eu-central-1, and ap-northeast-1 variants exist to match your account’s region.) Auth is either OAuth — which grants access across the workspaces in your billing account — or a bearer API token for headless use. Add the endpoint to Claude under Settings → Connectors as a custom connector; you’ll need a paid Claude plan for that, and the connector responds only inside chats you initiate.
Crucially, the official server is read-only. It exposes your workspaces, bots, calendars, and bot logs, plus Recall.ai’s documentation. That mix tells you its intended audience: engineers building on the platform. It makes Claude a genuinely good debugging companion:
- “Bot
bot_7f3a…never joined this morning’s Zoom call — pull its logs and tell me where the join flow failed.” - “List every bot that recorded a meeting yesterday and flag any that ended in an error state.”
- “Check the docs: what’s the right webhook to subscribe to for real-time transcript chunks?”
Because docs and live workspace data sit behind the same connector, Claude can correlate “what the API is supposed to do” with “what your bot actually did” — a real time-saver mid-integration.
What read-only rules out
Two things don’t fit through this pipe. First, dispatch: the official server can’t call Create Bot, so “send a bot to my 3pm” isn’t possible from a Claude chat unless you build and host a custom MCP server over the write endpoints yourself — feasible, since it’s a clean REST API, but now you own that code and its credentials.
Second, and more fundamental for the team-lead use case: the event flow points the wrong way. Recall.ai’s whole design is push — webhooks fire when a recording finishes or a transcript chunk arrives. Claude’s connectors cannot receive a webhook; nothing that happens in Recall.ai can open a Claude conversation. The moment a customer call ends and the transcript lands is exactly the moment you’d want action items extracted and the recap sent, and it’s exactly the moment nobody is sitting in a chat window prompting.
Pairing the API with an agent that catches events: Carly
If what you actually want is the downstream workflow — recording lands, action items come out, people get told — that’s Carly’s home turf. Carly is an AI executive assistant that runs trigger-driven workflows in the cloud: a finished recording can be the trigger, and the workflow runs whether your laptop is open or not.
You describe the pipeline conversationally — “when a Recall.ai recording finishes, pull the transcript, extract decisions and action items, post them in #sales, and email the recap to attendees” — and Carly interviews you on the details before building it with you. It sends real email (Gmail and Outlook), posts to Slack, and creates the tasks in Linear or Asana. AI agents start at $35/month, and steps in a workflow that don’t use AI run free and unlimited. See the Recall.ai integration page among Carly’s 200+ integrations.
Which tool for which half of the job
| Claude (official Recall.ai MCP) | Carly | |
|---|---|---|
| Inspect bots, calendars, and bot logs | Yes (read-only) | Yes |
| Debug a failed bot join against the docs | Yes — its best trick | Not the use case |
| Dispatch a bot to a meeting | No (read-only; needs custom MCP) | Yes, in a workflow |
| Acts when a recording finishes (webhook-style) | No | Yes |
| Sends recaps and files action items unattended | No | Yes |
| Setup | Hosted endpoint + OAuth/API key (paid Claude plan) | Guided conversation |
| Pricing | Paid Claude plan | AI agents from $35/mo |
Frequently Asked Questions
Does Claude integrate with Recall.ai?
Yes — Recall.ai hosts an official MCP server (regional endpoints like us-east-1.recall.ai/mcp) that you add to Claude as a custom connector with OAuth or an API token. It’s read-only and paid-Claude-plan-gated, and it shines for developers inspecting bots, logs, and docs.
Can Claude send a Recall.ai bot to record my meeting?
Not through the official server — it exposes no write operations. You’d have to build a custom MCP server around Recall.ai’s Create Bot endpoint and host it yourself. Inside a Carly workflow, bot dispatch and the follow-up are handled for you.
Can Claude process a transcript the moment a call ends?
No. Recall.ai signals “recording done” by webhook, and Claude connectors can’t receive webhooks or start conversations on their own. That handoff — transcript in, recap and tasks out, automatically — is what Carly is built around.
Is the Recall.ai MCP server official?
Yes, it’s built and hosted by Recall.ai and documented in their developer docs, with regional endpoints and both OAuth and bearer-token auth. As of mid-2026 there’s no Recall.ai app in Claude’s connector directory, so you add the endpoint manually.
Who is this integration actually for?
Primarily the engineers building on Recall.ai — it turns Claude into a live debugger with the docs attached. Teams that just want meeting recaps handled should skip the MCP wiring and set up a trigger-based agent instead; Carly’s AI agents start at $35/month.
More: Claude connectors · Can Claude send emails · Claude vs Carly · Claude Cowork alternatives · Claude meeting notes · Claude + tl;dv · Claude + Lever · Claude + Rippling
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
