AI meeting notetakers in 2026: how to pick after Fathom capped its free plan
Fathom's free plan now limits advanced AI summaries to a handful per month, which changes the default for small teams. The decision is not about feature lists, it is about meeting volume and privacy posture: solo or light users can stay free, while a team with more than a few client meetings a week is usually better served by a paid tier such as Fireflies. Pick on how much you actually meet, not on the comparison table.
Holding·reviewed30 May 2026·next+30dIf you use an AI meeting notetaker, you may have noticed your free tool got a catch this year. Fathom, which won a lot of small-business users on a famously generous free plan, now limits advanced AI summaries on its free tier to roughly five per month. The recording still works; the part most people actually want, the AI summary, is now metered. That is a small change with a clear consequence: it pushes the choice of notetaker back onto the table for a lot of small teams.
The useful way to make that choice is not to reopen the feature-comparison spreadsheet. It is to decide on two variables that actually determine the answer: how much you meet, and your privacy posture. This piece walks both, names the prices that matter as of May 2026, and ends with a 30-minute process to pick. Treat every price here as a starting point to verify on the live page, because notetaker pricing moves often.
What actually changed
Fathom’s appeal was the free tier. Through its growth it positioned free as effectively unlimited for everyday use, which made it the obvious default for solo founders and small teams. The 2026 change caps the advanced, templated summaries on free at roughly five a month; after that, free users still get unlimited recording, transcription, and a basic chronological summary, but the richer templated summaries are metered. For a light user, that cap may never be reached. For anyone running several client calls a week, it is reached early in the month, and for the rest of the month you have recordings and basic notes without the advanced summary that made the tool feel premium.
That is not a reason to be annoyed at Fathom; metering a costly AI feature on a free plan is a predictable move. It is a reason to re-decide, because the thing that made Fathom the no-brainer default for heavier users has changed. And the re-decision is genuinely simple once you frame it correctly.
The variable that decides it: volume
The mistake is to compare notetakers on features. They mostly do the same things: join a call, transcribe it, summarise it, let you search it. The variable that actually separates them for a small business is meeting volume crossed with seat count.
If you are solo or run only a handful of meetings a month, a free tier may still cover you, and a single-seat upgrade is a small cost if the cap bites. Fathom Premium is around $16 per month billed annually per its May 2026 pricing page, which is a reasonable single-user upgrade if you like the tool and just need the cap lifted.
If you are a team with several client meetings a week and more than one person needs their own notes, the math changes, and a per-seat team tool usually wins on value. Fireflies Pro is around $10 per user per month billed annually per its May 2026 pricing page, with a larger minute allowance and team features, which tends to make it the better default at that volume. Multiply by the seats you actually need to get the real monthly number; per-seat pricing looks cheap until you count the seats.
The decision, then, is almost mechanical: light and solo, stay free or take a single-seat upgrade; heavier and multi-person, move to a per-seat team plan sized to your volume. The feature list rarely changes that answer.
The privacy posture most operators skip
The second variable is the one small businesses tend to skip and sometimes should weigh first: what these tools do with the conversation.
A notetaker records and transcribes your calls, often by sending a bot that visibly joins the meeting. That raises three practical questions. There is consent: in many jurisdictions you should tell the other party they are being recorded, and a bot named after your company joining the call is itself a form of disclosure, while a quieter background capture may not be. There is data residency and handling: the transcript of a client call is sensitive, and you are trusting the vendor to store and process it appropriately, which is the same due-diligence question the vendor due-diligence piece applies to any tool that touches client data. And there is training: check whether your conversations are used to improve the vendor’s models, and whether you can turn that off.
This is why some users prefer a tool like Granola, which works more from your own notes than from always inserting a bot, and why Otter and others remain in the mix. None of this changes the volume logic; it adds a filter that can rule a tool out before price matters. For client-confidential work, the privacy answer can outrank the summary quality.
The Saturday decision
The 30-minute version, in order, is the how-to box above this section: count your real weekly meeting load and the seats that need notes; set your one-line privacy rule; check the two current prices against your volume on the live pricing pages; and trial your two finalists for a week of real meetings before committing to one and cancelling the other.
The only discipline that matters is judging the finalists on the transcripts and summaries you actually get in a week of genuine use, not on the comparison table. The stack-consolidation piece makes the broader version of this point: the lean stack is one tool per job, chosen on real use, not a drawer of overlapping subscriptions. A notetaker is one job; pick one tool for it.
What changes this recommendation
The cadence here is 30 days, with a review on 29 Jun 2026, for one reason: pricing and free-tier limits in this category move frequently, and the numbers in this piece are May 2026 list prices that may not hold.
The specific things to re-check: whether Fathom’s free cap moves, whether Fireflies changes its per-seat price or minute allowance, and whether a new entrant resets the value calculation at the bottom of the market. The decision logic, choose on volume and privacy, not features, is durable; the prices attached to it are not. Re-check both pricing pages before the review date and treat any figure here as a number to confirm rather than a current quote.
Related reading
For the wider question of how few tools a solo operator actually needs, see the stack-consolidation piece. For handing repetitive inbox and admin work to AI alongside your notetaker, see the email-triage stack and what to delegate to AI. For checking any vendor that will hold client data, see vendor due diligence in one Saturday. The claim behind this piece is tracked at its Holding-up entry.
Spotted an error? See corrections policy →