Fathom
AI meeting assistant that records, transcribes, and summarizes meetings
Verdict
Common use cases
- Pull weekly traffic summaries for standups
- Compare landing page performance across campaigns
- Identify top referral sources before ad buys
- Check real-time visitor counts during launches
- Export monthly analytics for client reports
Integration
- Vendor
- Fathom
- Category
- other
- Auth
- OAUTH2
- Composio slug
fathom
Tools
Setup
Setup guide
- 11. In Switchy, open your workspace settings and navigate to the Integrations tab. 2. Find Fathom in the MCP directory and click Connect. 3. You'll redirect to Fathom's OAuth consent screen — sign in with your Fathom account credentials. 4. Grant read access to your site analytics data (Fathom requests view-only permissions by default). 5. Switchy confirms the connection and returns you to the integrations page. 6. Open any Space and type '@Fathom show me last week's top pages' to test — the MCP should respond with a summary of your highest-traffic URLs. 7. If you manage multiple sites in Fathom, specify the site ID in your prompt or set a default site in the MCP settings panel.
What teammates see: by default, memories from Fathom are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Spaces have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Weekly Traffic Report
@Fathom show me total visits, page views, and average session duration for the last 7 days. Include the top 5 pages by traffic.Open in a Space →
Referral Source Breakdown
@Fathom list the top 10 referral sources for this month, with visit counts and percentage of total traffic for each.Open in a Space →
Landing Page Performance
@Fathom compare bounce rate and average time on page for /pricing, /features, and /demo over the last 30 days.Open in a Space →
Real-Time Visitor Check
@Fathom how many visitors are on the site right now, and which pages are they viewing?Open in a Space →
Monthly Growth Snapshot
@Fathom show me total visits and goal completions for this month versus last month, with percentage change for each.Open in a Space →
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@Fathom show me the top 5 pages by traffic for the last 30 days on our marketing site
Based on your Fathom analytics, here are the top 5 pages by pageviews over the last 30 days: 1) /pricing (8,432 views), 2) /features (6,201 views), 3) /docs/getting-started (4,987 views), 4) /blog/launch-announcement (3,654 views), 5) /integrations (2,891 views). Total unique visitors across these pages: 14,203. The pricing page saw a 23% increase compared to the previous 30-day period.
This example assumes read access to site analytics via OAuth. Fathom's privacy-first approach means no personally identifiable visitor data is available — only aggregate metrics like pageviews, referrers, and device types. Useful for quick traffic snapshots without opening the dashboard.
@Fathom what's our bounce rate and average time on site for mobile visitors this week compared to desktop?
For the current week: Mobile visitors have a 62% bounce rate with an average session duration of 1m 47s. Desktop visitors show a 48% bounce rate with an average session duration of 3m 12s. Mobile traffic represents 41% of total visits (2,834 sessions), while desktop accounts for 59% (4,102 sessions). Mobile bounce rate is up 5 percentage points week-over-week, suggesting potential mobile UX issues worth investigating.
Fathom tracks device type and engagement metrics without cookies or fingerprinting. This comparison helps identify platform-specific performance gaps. Note that 'bounce rate' in Fathom means single-page sessions with no additional interactions — definition may differ from Google Analytics.
@Fathom summarize referral sources driving traffic to our docs section over the last quarter and suggest which partnerships to prioritize
Top referral sources to /docs/* over the last 90 days: 1) github.com (1,847 visits), 2) dev.to (923 visits), 3) reddit.com/r/webdev (612 visits), 4) hackernews (487 visits), 5) twitter.com (301 visits). GitHub referrals convert to sign-ups at 18%, significantly higher than other sources (avg 7%). Recommendation: Prioritize GitHub presence — consider sponsoring popular repos in your ecosystem, improving README examples, and engaging in issue discussions. Dev.to shows strong volume but lower conversion; A/B test more technical deep-dives there.
This synthesis combines Fathom's referrer data with the AI's strategic reasoning. Fathom provides the traffic sources and volumes; conversion rate analysis would require correlating with sign-up data from another system. The AI can spot patterns but can't access user-level journeys due to Fathom's privacy model.
Use-case deep-dives
When recorded standups replace live sync for distributed teams
A 6-person engineering team across three time zones records their standups instead of scheduling a live call. Fathom transcribes and summarizes each recording, then surfaces action items and blockers in a shared workspace. This works when the team already uses video standups and needs structured outputs—transcripts, speaker labels, and keyword extraction—without manual note-taking. The OAuth2 flow means each team member authenticates once, and Switchy can pull meeting data into shared context for sprint planning or retrospectives. The trade-off: if your team prefers text-first async updates in Slack or Linear, adding a video layer introduces friction. This MCP wins when video is already the medium and you need the notes automated, not when you're trying to force video into a text-native workflow.
Building searchable call archives for support and sales handoffs
A 3-person customer success team takes 15-20 calls per week with enterprise customers. Fathom records and transcribes each call, and Switchy indexes the transcripts so the team can search past conversations when a customer escalates or a sales rep needs context before a renewal. This scenario depends on the MCP exposing search or retrieval tools—without them, the integration is just a one-way archive. The OAuth2 setup means each rep's calls stay scoped to their account unless explicitly shared. This works for teams that already record calls and need a central search layer, not for teams starting from scratch with call recording policies. If your call volume is under 10 per week, a shared Google Doc might be faster than wiring up an MCP.
When recorded onboarding sessions become living documentation
A 10-person startup runs weekly onboarding sessions for new hires, covering product architecture, customer workflows, and internal tools. Fathom transcribes each session, and Switchy compiles the transcripts into a searchable onboarding wiki that updates as the product evolves. This scenario assumes the MCP can export or stream transcript data—if it only stores recordings in Fathom's app, the integration is a dead end. The win is when onboarding is already video-based and the team wants to retire static docs that go stale. The threshold: if your onboarding is under 3 sessions or already text-native in Notion, the overhead of video transcription doesn't pay off. This MCP fits teams that record training anyway and need the content repurposed, not teams looking to start recording just for documentation.
Frequently asked
What does the Fathom MCP do in Switchy?
The Fathom MCP connects your team's Fathom Analytics account to Switchy's AI workspace, letting you query site traffic, page views, and visitor metrics without leaving your chat. Since Fathom is privacy-first analytics, you get clean data without cookie banners or GDPR headaches. Your AI can pull reports, compare time periods, and answer questions like 'which pages drove the most conversions last week' on demand.
Do I need admin access to connect Fathom via OAuth?
Yes. Fathom's OAuth flow requires you to authorise Switchy with an account that owns or has admin rights to the sites you want to query. If you're a read-only team member in Fathom, the connection will fail or return empty data. The person who connects it controls which sites Switchy can see, so coordinate with your analytics owner before linking.
Can the Fathom MCP send data or change site settings?
No. Fathom's API is read-only by design — the MCP can fetch aggregated stats, event counts, and referrer breakdowns, but it cannot edit goals, delete sites, or inject tracking code. If you need to change Fathom configuration, you still do that in the Fathom dashboard. This MCP is purely for pulling insights into Switchy's AI context.
How is this different from just logging into Fathom directly?
Logging into Fathom gives you charts and filters; the MCP gives your AI the raw numbers to reason over. You can ask 'compare mobile vs desktop traffic for the last two launches' and get an answer in seconds, without building custom dashboards. It's faster for ad-hoc questions and works alongside your other tools in Switchy — your AI can correlate Fathom data with GitHub deploys or Slack feedback in one thread.
Who on the team should connect the Fathom MCP?
Whoever owns your Fathom account or has admin access to the sites you want to analyse. Once connected, anyone in your Switchy workspace can query the data through AI — they don't need their own Fathom login. If you rotate analytics ownership, you'll need to reconnect the MCP with the new admin's OAuth token to keep it working.