How to Connect Fathom to Claude (And Turn Meeting Notes Into Actionable Intelligence)
You’re paying for Fathom. Your meetings are recorded. Your transcripts are generated. Your AI summaries sit in a dashboard.
And then what?
You manually search through recordings when you need to find that one thing a client said. You copy-paste action items into your task manager. You forget half the follow-ups because they’re buried in a summary you read once and never opened again.
Here’s the thing: Fathom is great at capturing meetings. But it wasn’t built to analyze them across time, spot patterns in your behavior, or let you query your call history like a database.
That’s exactly what happens when you connect Fathom to Claude using the Fathom MCP Server.
We’ve been running this setup at onPoint Studio for the past few months, and it changed how we handle everything — from client follow-ups to sales coaching to internal retrospectives. Here’s how to set it up and what you can actually do with it.
Fathom Now Has an Official MCP Connector (April 2026)
Fathom released a native, first-party MCP integration on April 25, 2026. If you’re starting fresh, this is the faster path — no custom URLs, no third-party server.
For Claude
- Go to claude.ai/directory/connectors/fathom
- Click Connect and complete the authorization prompts
- Done — ask Claude about your meetings directly
For other MCP-compatible tools, use api.fathom.ai/mcp.
The official connector gives you the same core capabilities — query meetings, pull transcripts, draft follow-ups — without any setup overhead.
Prefer more control? The third-party fathom-mcp-server.com connector covered below still works and remains the better option if you want self-hosting, a custom deployment, or need to understand the full data flow. The rest of this guide covers that path.
What Is MCP (And Why Should You Care)?
MCP stands for Model Context Protocol. Think of it as a standardized way to connect AI models like Claude to external data sources — your calendar, your CRM, your meeting notes.
Without MCP, Claude only knows what you paste into the chat. With MCP, Claude can pull data directly from Fathom: your meetings, transcripts, summaries, action items, and team info. No copy-pasting. No context switching.
The Fathom MCP Server is an open-source connector (MIT license) that acts as a secure bridge between your Fathom account and Claude. It’s a pass-through — your data flows from Fathom to Claude without being stored, cached, or logged anywhere in between.
How It Works: The Architecture
The setup is simple. Three components:
- Fathom AI — where your meetings, transcripts, and summaries live
- MCP Server — the secure proxy that handles OAuth authentication and exposes 6 API tools to Claude
- Claude — where you query, analyze, and act on your meeting data
The MCP server uses OAuth 2.0 for authentication, encrypts tokens with AES-256-GCM, and operates over HTTPS only. Your Fathom data is never stored on the server — it’s fetched on demand and passed directly to Claude.
Setup: 60 Seconds, No Code
This is the part that surprised us. There’s no npm install, no API keys to manage, no server to deploy. The hosted version handles everything.
Step 1: Open Claude Settings
In Claude (desktop app or claude.ai), go to Settings → Connectors → Add Custom MCP.
Step 2: Paste the MCP Server URL
Enter this URL:
https://www.fathom-mcp-server.com/mcp
Claude will open a Fathom OAuth popup. Click Allow to grant read-only access to your meeting data.
That’s it. The connection is established.
Step 3: Start Querying Your Meetings
Open a new conversation in Claude and start asking questions about your calls. Claude now has access to 6 Fathom tools:
list_meetings— browse meetings with date, team, and recorder filterssearch_meetings— find meetings by title or keywordget_transcript— pull the full transcript of any recordingget_summary— get the AI-generated summary with topics, takeaways, and next stepslist_teams— see all teams you have access tolist_team_members— view members of any team
For Organizations
If you’re on a Claude team or enterprise plan, admins need to add the connector through organization admin settings — not personal settings. This makes the Fathom integration available to everyone on the team.
What You Can Actually Do With It
Here’s where it gets interesting. Listing meetings is table stakes. The real power is in what Claude can do with your meeting data once it has access.
1. Query Meetings Like a Database
Instead of scrolling through a list in Fathom’s dashboard, just ask:
- “Show me all meetings with [client name] this month”
- “What did we discuss in last Tuesday’s standup?”
- “Find the call where we talked about the payment gateway options”
Claude will search, filter, and return exactly what you need.
2. Cross-Meeting Analysis
This is the feature that doesn’t exist anywhere else. Claude can pull summaries from multiple meetings and synthesize them:
- “Compare what the client said in the intro call vs the proposal call — did their requirements change?”
- “Across my last 10 client calls, what’s the most common objection?”
- “Show me all action items assigned to me in the past 2 weeks”
3. Behavioral Analysis
We ran this on our own call history at onPoint Studio. Claude analyzed 20 calls across 3 months and surfaced patterns we’d never noticed:
- 70% of our calls were client-facing — revealing a single-point-of-contact bottleneck
- Most “15-minute” calls were running 30-40 minutes — a consistent overrun pattern
- Action items were heavily skewed toward operational tasks (“email X to Y”) vs strategic work
That kind of insight doesn’t come from reading individual summaries. It comes from giving an AI access to the full dataset and asking the right questions.
4. Meeting Prep
Before a follow-up call, ask Claude: “Summarize everything we’ve discussed with [client] so far, including all open action items.” You walk into the call fully briefed — without spending 20 minutes re-reading old notes.
5. Sales Coaching
Pull transcripts from sales calls and ask Claude to analyze your pitch patterns, objection handling, or how effectively you’re qualifying leads. We used this to build a documented 4-stage sales process for our team.
Security: What You Need to Know
It wouldn’t connect client meeting data to anything without checking the security model first. Here’s what is verified:
- Pass-through architecture — the MCP server fetches data from Fathom and passes it to Claude. Nothing is stored, cached, or logged.
- OAuth tokens only — the only data the server persists are encrypted OAuth tokens (AES-256-GCM).
- HTTPS everywhere — all communication between Claude, the MCP server, and Fathom is encrypted in transit.
- Read-only access — the Fathom API only exposes read operations. The MCP server can’t modify your meetings or data.
- Open source — the entire codebase is on GitHub under MIT license. You can audit it yourself.
Self-Hosting (Optional)
If your organization requires full control over the data flow, you can self-host the MCP server. The project runs on Node.js 18+ with a PostgreSQL database. The recommended deployment path is Railway, but any platform with Node.js, PostgreSQL, and HTTPS will work.
The self-hosting setup involves:
- Forking the GitHub repo and deploying to your hosting provider
- Creating a Fathom OAuth App in the Fathom Developer Portal
- Setting environment variables (database URL, encryption key, OAuth credentials)
- Running database migrations
- Pointing Claude at your custom URL instead of the hosted one
Full instructions are in the project README.
What’s Coming Next
The project roadmap includes some features we’re excited about:
- Transcript vectorization — better parsing of long transcripts for more accurate AI analysis
- Action item aggregation — cross-meeting action item tracking with filters
- Meeting analytics — total meeting time, frequency, and top attendees
- Speaker time analysis — who talks most in each meeting
The project is actively maintained and open to contributions.
The Bottom Line
If you’re already using Fathom and Claude, this integration takes 60 seconds and unlocks a completely different way of working with your meeting data. Instead of passively reading summaries, you’re actively querying, analyzing, and acting on everything your team discusses.
We use this daily at onPoint Studio — for client follow-ups, sales coaching, internal retrospectives, and behavioral analysis. It’s one of those setups where the value compounds over time as your meeting history grows.
The connection URL again:
https://www.fathom-mcp-server.com/mcp
Try it. Pull your last 10 calls. Ask Claude to find something you forgot to follow up on. You’ll see what we mean.
Need help building AI-powered workflows into your WordPress or WooCommerce setup? This kind of tool integration is exactly what we do at onPoint Studio. Book a free intro call and let’s talk about what’s possible.


