AI Notetaking Apps 2026: Use-Case Picks Collide With Consent Lawsuits

A buyer's guide that shows how the market has split

On October 3, Metaview published its list of the 13 best AI notetaking apps for 2026. It does not name one winner. Instead it argues that the right tool depends on the job. Its picks are its own Metaview Notetaker for recruiting calls, Gong or Avoma for sales, Dovetail for user research, Fireflies or Fathom for general meetings, and the recaps built into Microsoft Teams or Zoom for teams that only need the basics.11 The full list also includes Granola, Shadow, Sembly AI, tl;dv and Grain. Prices run from free tiers up to quote-only enterprise contracts.11

The publisher is not neutral. Metaview sells a recruiting product and puts it first. That bias is worth discounting, but the guide's structure is still useful. It shows that the AI meeting assistant has stopped being one product and become several. Read next to other 2026 roundups, the use-case approach looks right. What most of these lists leave out is the legal risk now hanging over the category, and that risk matters at least as much as price or features.

Comparison guides agree on use cases, not on the details

The roundups mostly agree on who does what. Fathom keeps coming up as the best free option. Several guides point to its unlimited free recording, and one ranks it first overall on G2 rating and how quickly it produces summaries.212930 Fireflies is the usual pick for teams that need CRM integrations and many languages, starting at about $10 per user per month on annual billing.232728 Granola is the standard answer for people who don't want a bot visibly joining their calls.252730 Gong stays the enterprise choice for revenue intelligence. Metaview reserves it for sales teams of roughly 25 reps or more.11

One notable gap: Otter.ai, one of the best-known names in the category, is not among Metaview's 13.11 Many competing guides still include it, usually as the pick for live transcription and captions.222930

Pricing is where the guides disagree, and buyers should pay attention to that. Metaview's own entry price varies a lot depending on who you ask. One recruiting-focused comparison puts Metaview Pro at $100 per user per month and says the company now charges per agent across a wider platform.12 Another lists it at about $50 to $60 a seat.15 A third cites $60.13 A review published in August explains part of the gap. Metaview is rebranding itself as an "Agentic Recruiting Platform," with sourcing tiers at $100 and $300 per user per month, while the full agent bundle is priced by quote.14 Granola's numbers also differ, from $14 per user per month in several guides to $18 in Metaview's table and others.112322 Some of this comes from monthly versus annual billing and different plan names. It still means a list price in any roundup should be treated as an estimate, not a quote.

Notetakers are becoming agents

The bigger shift is in what these products are trying to be. Metaview built its name as a recruiting notetaker and is now selling notetaking as one piece of a broader agentic platform.12 One reviewer says the notetaker is still the best-reviewed part of the product, with a 4.8 out of 5 rating on G2 across 131 reviews. The same reviewer says buyers should look most carefully at the newer metered sourcing features and the custom-priced platform.14

Competitors are moving the same way. One vendor promotes post-meeting follow-up, with drafted emails and extracted action items.21 Another sells live coaching during the conversation itself and notes that Otter, Fireflies, Zoom AI Companion and TwinMind now offer some in-call help too.24 In my view, the transcript has become a commodity, and vendors are trying to compete on what they do with it.

This also blurs the line between meeting assistants and AI writing tools. A separate group of products treats notes as material for synthesis rather than as a meeting record. Guides focused on that group recommend Notion AI as an all-in-one workspace, Obsidian for people who want local files they control, and NotebookLM for research grounded in your own documents with citations.1920 Zapier's list leans toward the writing side, praising Evernote's AI editing tools for fixing typos, summarizing and changing tone.18 Meeting assistants that draft follow-ups and update other systems are now moving into that territory from the other direction.

The bot-free trend is now a legal problem

The most important context this year is legal. Over the past year, AI notetakers have gone from a convenience to a subject of litigation. As of mid-October 2026, suits were pending against Otter.ai, Fireflies.ai, Granola and Microsoft over Teams live transcription.2

The Otter case is the furthest along. Four proposed class actions filed in August and September 2025 were combined before Judge Eumi K. Lee in the Northern District of California.8 The plaintiffs say Otter's meeting bot joins calls automatically through calendar integrations, records people who never created an account, builds voiceprints of speakers, and uses de-identified recordings to train its models.2 Otter argued that consent was the account holder's responsibility and that no interception took place.7 On August 13, Judge Lee let the claim under section 631 of the California Invasion of Privacy Act move forward. She found the plaintiffs had plausibly alleged that Otter collects, keeps and uses conversations for its own commercial purposes, which could make it a third-party eavesdropper.5

That reasoning has consequences for the whole category. It shifts the question from whether a bot is visible to what the vendor does with the audio afterward. Fireflies faces a related but different problem: suits under Illinois' biometric privacy law over voiceprints created by its speaker-recognition feature.3 One legal analysis points out that recording a meeting and building a biometric database from the people in it are two different acts.3 Microsoft Teams faces a similar biometric class action in Illinois.6

The Granola case undercuts the main selling point of the bot-free approach. Chamberlain v. Granola, Inc., filed July 30 in the same federal court, alleges the app was deliberately designed to record without telling everyone on the call. Granola captures audio directly from the user's computer instead of joining as a participant.1 According to one summary of the allegations, Granola's optional notice features are off by default. The complaint also says the company marketed the product by telling users that others in the room "won't know it's there," and that it uses anonymized data to improve its models by default on Free and Business plans.5 None of this has been proven, and no court has ruled on the Granola complaint.3

The guides do not frame this the same way. Meeting-assistant roundups still recommend Granola as the top pick for people who dislike bots.2729 Legal commentary makes the opposite point: with no visible bot, participants get less notice, not more.29 Both can be accurate. The roundups describe how the product feels to use. The lawsuits are about the people on the other end of the call who never chose to be recorded.

What buyers should take from this

The exposure is real money. One compliance guide counts 12 states that require all-party consent to record. It also notes that a 2006 California ruling, Kearney v. Salomon Smith Barney, can apply the state's law whenever a Californian is on the call.10 Penalties can reach $5,000 per violation under California's privacy law and under Illinois' biometric law for intentional violations.106 For Metaview's main audience, recruiters, the risk is direct. One employment lawyer has said HR teams should watch the Otter case closely, since recording a candidate with a tool that captures voiceprints could create biometric-law exposure whether or not the candidate is hired.10 Metaview's competitors stress that it does not train its models on customer data and holds SOC 2 Type II certification, and that matters more now than it did a year ago.15

The built-in recaps at the bottom of Metaview's list may be the safer default for many companies. One October comparison found that Zoom shows every participant a prompt, Google Meet displays a visible indicator, and Teams can require per-attendee consent when an admin turns on the policy. Otter, by contrast, still lists training on de-identified audio.5 Choosing tools by use case is still sensible. My reading is that consent should come before features in any 2026 purchase. Teams can no longer just ask which notetaker fits the workflow. They also need to ask whether everyone on the call knows it's running and what the vendor does with the recording afterward.