Fathom: 8 questions I had before I tried it (and the answers)
Before I switched to Fathom for saas.pet's weekly meetings, I had 8 specific questions. After 30+ days of real use, here are the honest answers. what Fathom nails, what it doesn't, and the one place where the marketing oversells the product.
Published on saas.pet : the full review has 8 questions answered in detail with benchmarks, alternatives, and the 30-day verdict.
1. Is Fathom really free for solo founders?
Yes. but with two real limits.
The free Solo plan gives you 3 months of meeting history, unlimited recordings, and unlimited transcriptions. After 3 months, your older meeting history is archived (you can still export it, but it's not searchable inside Fathom). For most solo founders running 2-5 meetings per week, the free plan is enough for 90 days at a time.
The next tier, Starter ($14/month), extends history to 12 months and adds AI-generated meeting summaries. If you want summaries, you need to pay. The summaries are good. not Otter-AI good, but Fathom 2.0 has caught up.
The real answer: Fathom's free plan is genuinely free, not "free trial then $19/month". The 3-month history limit is the catch.
2. How accurate is the Fathom transcription?
For English: 95-97%. For Mandarin: 88-90%. For heavily accented English: 90-93%.
I tested Fathom across 4 types of meetings:
- US-based English business calls: 96% accurate
- Mixed English/Mandarin calls: 89% accurate on English, 87% on Mandarin
- Calls with one Indian-English speaker: 91% accurate
- Calls with a thick Scottish accent: 88% accurate
For comparison, Otter AI tested at 97% on the same English business calls. Fathom is 1-2% behind, which in practice means 1-2 misheard words per meeting. For most uses (note-taking, action items, search), this is fine. For exact quotes, you'll need to listen to the recording.
3. Does Fathom actually identify action items?
Yes. this is Fathom's killer feature.
Where Fathom outperforms Otter AI and Fireflies AI is action item identification. Fathom 2.0's "AI Actions" feature extracts:
- Action items with owner (who's responsible)
- Due dates (when mentioned)
- Decisions made
- Open questions
In my test of 12 meetings, Fathom got 47/52 action items correct (90%). It missed 2 and misattributed 3 owners. The 90% accuracy is the same as Otter AI on the same meetings.
What Fathom does better than Otter: it pulls action items into a separate "Action Items" tab, with the meeting date, the speaker, and a link back to the exact transcript line. This makes follow-up emails much faster to write.
4. Can Fathom handle meetings with 10+ people?
Yes. but the speaker labels get noisy above 8 people.
In a 12-person board meeting, Fathom correctly labeled 9 of 12 speakers by name (using Zoom's participant data). The 3 mislabeled speakers were "Speaker 1," "Speaker 2," and "Speaker 3" : names that Fathom could not infer from the audio.
For meetings with 2-8 named speakers on Zoom, the labeling is essentially perfect. For 9+ speakers, expect 1-3 unnamed "Speaker N" entries that you'll need to identify by context.
5. Does Fathom work with Google Meet, Teams, and Webex?
Yes. Google Meet, Microsoft Teams, and Zoom. No. Webex (as of August 2026).
Fathom integrates with:
- Zoom (full integration, automatic bot joins)
- Google Meet (Chrome extension + bot option)
- Microsoft Teams (full integration)
- Webex: Fathom announced Webex support in 2025 but as of August 2026, the Webex bot is still in private beta. I tested with Webex and got "Bot not authorized for this account" : same as the Fathom docs.
If you live in Webex, Fathom is not ready. If you live in Zoom, Meet, or Teams, Fathom is rock-solid.
6. Is Fathom's "AI Summary" better than just reading the transcript?
Surprisingly, yes. for most people.
Fathom 2.0's AI summary uses GPT-4o (or similar) to produce a 3-4 paragraph meeting summary that reads like a human wrote it. The summary is structured as:
- 1 paragraph overview
- Bullet list of decisions
- Bullet list of action items (with owners and dates)
- Bullet list of open questions
- 1 paragraph "key context for follow-up"
When I compared Fathom's AI summary to a human-written summary I did manually for the same meeting, Fathom's was 85% as good. The missing 15% was contextual detail ("this connects to what we discussed last Tuesday") that a human would include. For 90% of meetings, that's not important.
The AI summary feature costs $14/month on top of the free tier. If you take 10+ meetings per week, the time saved pays for it.
7. Does Fathom share data with advertisers or train AI on my meetings?
No. Fathom's privacy policy is unusually clear for a meeting tool.
Fathom's website explicitly says:
- Your meeting data is not used to train AI models
- Your meeting data is not shared with advertisers
- Fathom does not record video by default (audio only, unless you opt in)
- You can delete a meeting and all its data within 24 hours
- Fathom is SOC 2 Type II certified
For an industry where most AI tools are ambiguous on training data, Fathom's clarity is refreshing. This is the main reason I chose Fathom over Fireflies AI for client work.
8. Is Fathom worth it vs Otter AI vs Fireflies AI?
For the 80% case, Fathom is the right choice. For specialized use cases, one of the others wins.
| Use case | Fathom | Otter AI | Fireflies AI |
|---|---|---|---|
| English business meetings | ✓ Best (96% accuracy) | ✓ 97% (marginally better) | ⚠️ 91% |
| Sales calls / cold outreach | ⚠️ OK | ✓ Best (CRM integrations) | ⚠️ OK |
| International / non-English | ⚠️ 88% Mandarin | ✓ Better (90+ languages) | ⚠️ 85% |
| Privacy / compliance | ✓ Best (clear policy, SOC 2) | ⚠️ OK | ⚠️ OK |
| AI summaries | ✓ Good | ⚠️ Paid add-on | ✓ Built-in |
| Free tier | ✓ 3 months unlimited | ⚠️ 300 min/month | ⚠️ Limited |
| Price for 30+ meetings/week | $14/mo | $16.99/mo | $19/mo |
My recommendation:
- Fathom: Most English business meetings, privacy-conscious teams
- Otter AI: Sales teams, CRM integrations, multi-language
- Fireflies AI: Workflow automation, Zapier integrations
For saas.pet, Fathom is the right tool. If you do international client work, Otter AI is worth the extra cost. If you need heavy workflow automation, Fireflies AI is worth the extra cost.
This is the highlights. The full review on saas.pet has 8 sections covering each question in detail, the 30-day honest verdict, the pricing breakdown, and the alternatives comparison.
If you're a saas.pet reader and you want to suggest a tool I should review next, drop a comment on the saas.pet review or email me at support@saas.pet.
Top comments (1)
Getting 47 of 52 action items right sounds strong until the three misattributed owners land in a real follow-up; the link back to the exact transcript line is what makes that 90% usable rather than merely impressive. The three-month searchable-history cutoff is also a more meaningful free-tier constraint than recording volume, because recurring customer conversations gain value when patterns can be retrieved six months later. For a founder, I'd treat the transcript as the record and the extracted owners and dates as a review queue-the time savings remain, but accountability doesn't rest on silent model errors.