AI Meeting Assistants Split Rankings as Privacy Lawsuits Mount

Buyer's guides that don't agree on much

AI meeting assistants have become a crowded software category, and the buyer's guides published this year disagree about who leads it. In February, Reclaim.ai's list of the top 18 AI meeting assistants put Zoom AI Companion first. Its own scheduling product came second, labeled "best overall," followed by Gong, Avoma, Notion AI, Otter.ai, Fireflies.ai, tl;dv, Fathom, Krisp and Granola. The list also reached into calendar and email tools such as Vimcal, Superhuman and Calendly's waitlisted Notetaker.26 Rankings published since then have reshuffled that order again and again. TechRepublic's September guide names Fathom the best overall note taker, pointing to its free plan and its mix of bot and bot-free capture. Albato ranks Avoma first because it pushes call outcomes into the CRM. A review on MeetingNotes.com names Fellow as its top pick.24

The split partly reflects who writes these lists, and readers should weigh that heavily. Many of the most detailed rankings come from vendors that put themselves near the top. Reclaim does it.26 So does Fellow, in its 26-tool guide.25 Laxis calls itself the best note taker after testing five apps across more than 200 meetings.23 Convo says it is the best choice for most people.27 Cirrus Insight leads its own 13-tool comparison.22 The useful information in these guides sits in the shared middle. Nearly every list includes Otter, Fireflies, Fathom and Granola, and the labels they get are fairly consistent: Otter for live transcription, Fireflies for searchable archives and CRM integrations, Fathom for a generous free tier, and Granola for bot-free personal notes.2225

The real contest is what happens after the call

The clearest trend in this year's coverage is that transcription alone no longer sets products apart. A Simular review of eight tools says most stop at transcripts and summaries, and argues the winning product will be the one that turns notes into actions.21 Albato ranks its whole list by what happens after the call ends, and calls the gap between a clean transcript and an updated CRM "where most note-takers stop." A Luminix analysis argues that "good enough transcription plus summary" is finished as a product. In its reading, each survivor is betting on a different extra layer. Granola bets on humans and AI writing notes together, Fathom on free access with little friction, Fireflies on downstream automation, and Otter on real-time collaboration.3

That is why these tools now overlap with AI writing assistants and research copilots. Fellow's comparison tracks whether each product exposes meeting data through Model Context Protocol (MCP) servers or agent connectors. It reports that Fathom has an official MCP server with Claude and ChatGPT connectors on every plan, and that tl;dv offers MCP, an API and webhooks on its Pro tier.25 Granola launched its own MCP server on February 4, 2026. It also rebuilt its chat feature as an agent with inline citations in April, and added pre-meeting "Briefs" in May.9 Laxis sells itself partly as a voice keyboard for dictation between meetings, which makes it a writing tool as much as a recorder.23 Read together, these moves show the meeting record turning into a searchable store of company knowledge that AI assistants can query. The note taker becomes the intake layer for a research copilot.

Bot versus botless is no longer a clean line

For most of 2025, buyers sorted products by whether a visible bot joins the call. Simular's review still uses that split. It recommends bot-based tools such as Otter and Fireflies for internal standups, and botless tools such as Granola, Fathom and Krisp for client calls, sales meetings and HR conversations.21 The coverage now suggests the line is blurring. One reviewer notes that Otter and Fireflies have added bot-free desktop modes next to their meeting bots.2 Fellow's comparison reports that Fireflies added native Google Meet capture in early 2026. It also says Fathom has been botless on Google Meet only since April 2026, while still using a visible bot on Zoom and Teams.25

The reports don't fully match on this point. TechRepublic simply credits Fathom with "bot and bot-free capture." A Convo comparison still lists Fathom as having a visible bot.27 Fellow's platform-by-platform breakdown is the most specific of the three.25 The practical lesson is that "botless" now describes particular configurations of a product. It is no longer a stable feature buyers can count on.

Pricing is just as hard to pin down. The same tools show different entry prices depending on the source and billing period. Otter is listed at $16.99 a month in some guides and at $8.33 billed annually in others.21 Laxis argues that the headline free-tier numbers mislead. It points out that Fathom's unlimited recording comes with only five AI summaries a month, and that Otter's 300 free minutes are capped at 30 minutes per conversation.23

Granola's momentum and Google's entry

Granola gets most of the attention this year. Simular calls it the "2026 darling" of AI note-taking.21 Spending data backs that up. YipitData reports that Granola raised a $125 million Series C in March 2026 at a $1.5 billion valuation.4 Its transaction panel showed 3x spend growth over six months, churn near zero, and a pattern where customers dropped Fathom, Otter or Fireflies after adopting Granola, but rarely the other way around.7 YipitData also notes that Fireflies, Otter and Fathom still have larger customer bases. In its data, though, their spending has plateaued or declined.7

Competitors are now copying Granola's design. On October 8, TechCrunch reported that Google released AI Edge Foresight, an offline Mac app. It runs on-device models to transcribe meetings and generate notes, and uses the same split-screen layout of handwritten shorthand next to AI notes.10 A free, local-first option from Google puts pressure on startups that charge for the same workflow. Its offline design also suggests that keeping data on the device is becoming a selling point in its own right.

The lawsuits change the meaning of "private"

The most important development is one that most ranking lists leave out: the legal exposure. Three of the best-known products now face privacy lawsuits.5 Otter is the defendant in a consolidated case in the Northern District of California. On August 13, 2026, the judge granted only part of Otter's motion to dismiss. That let the federal Wiretap Act, California Invasion of Privacy Act and Illinois biometric claims go forward to discovery, on the theory that Otter may have acted as a third-party eavesdropper.5 The allegations say Otter asked only the meeting host for permission, and that other participants could not turn the tool off.19

Fireflies faces suits under Illinois's Biometric Information Privacy Act (BIPA). They allege that its speaker-recognition feature creates voiceprints without written notice or consent.1711 Forbes adds that Microsoft Teams faces a similar BIPA class action over its transcription feature.14

The case that changes the bot-versus-botless debate is Chamberlain v. Granola, filed July 30, 2026. It alleges that Granola recorded a participant who had no notice an AI note taker was present. It also alleges that meeting content was used for AI training by default unless the user opted out.2016 Plaintiffs cited Granola's own marketing line that other participants "won't know it's there."5 Lifehacker notes that only the person who installed the app can turn it off.11

The reports agree on the legal stakes and differ mainly on emphasis. Several commentators argue that removing the visible bot may make consent harder to prove, not easier.1316 One legal analysis concludes that the main issue was never the bot. It was consent, and what happens to the recording afterward.5 All of these cases are still unresolved, and none of the allegations has been proven.156

What it adds up to

The best way to read the 2026 market is that competition on features is now shaped by risk. Roughly a dozen U.S. states require all parties to consent before a conversation is recorded. California allows penalties of up to $5,000 per violation.13 For lawyers, a February 2026 ruling in US v. Heppner raises the risk that sharing conversations with cloud AI services could undermine privilege claims.18 Analysts also point out that using the tool yourself can put the consent obligation on you, not only on the vendor.1615

That changes what a good top-18 list should measure. Buyers should look past which tool writes the cleanest summary. The more important questions are whether every participant can see and decline the recording, whether the vendor trains AI on meeting content by default, and whether administrators can control how long data is kept. Fellow's comparison is notable for scoring retention controls, and it finds big differences across vendors.25 Industry voices increasingly treat consent as a design choice.1220 The products that come out ahead are likely to be the ones that add clear notice and default-off training to their agent features. Bot-free capture on its own no longer counts as a privacy guarantee.