Six AI notetakers are fighting over the most boring, most valuable half hour of your day. Otter, Fireflies, Granola, Fathom, Read AI, and Gumdrop each made a different bet on what a meeting really is.
Somewhere on your calendar right now there is a 30-minute block that nobody wants. A status sync. A kickoff. A call that could have been an email and became a meeting anyway. For most of software history, the memory of that half hour depended on one exhausted person half-typing while half-listening. Six companies decided that was a bad system, and they are now worth billions for saying so.
The category is called AI notetakers, which undersells it the way "horseless carriage" undersold the car. Otter, Fireflies, Granola, Fathom, Read AI, and Gumdrop all record what happens in a meeting and hand you back something useful. That much they share. What separates them is a much more interesting question: what do they think a meeting actually is?
Start with the two that got there first. Otter.ai and Fireflies.ai both launched in 2016, when transcription was a novelty you demoed at a party, not a tool you trusted with a board meeting. Otter came out of Mountain View, founded by Sam Liang and Yun Fu under the corporate name AISense. Fireflies came out of San Francisco, founded by Krish Ramineni and Sam Udotong, an MIT graduate who built the first version around a friendly bot the company calls Fred.
Both chased the same thing: scale. Otter now reports more than 35 million users and over a billion meetings processed, and in late 2025 it announced it had crossed $100 million in annual recurring revenue. It also did something the others mostly do not: it built its own transcription engine in-house rather than reselling someone else's, then tuned it for accents and languages that trip up cheaper systems. Otter's pitch has quietly shifted from "we transcribe your calls" to "we are your company's memory," and in 2025 it started shipping autonomous voice agents that can actually speak in a meeting, not just listen.
Fireflies took the more surprising financial route. It reached a valuation around $1 billion while raising only about $19 million total, a ratio that makes venture capitalists either very happy or very nervous. Its edge is reach: Fred drops into Zoom, Meet, and Teams, files everything into a searchable archive, and pipes the results into the CRM tools that sales teams live inside. When a company says three-quarters of the Fortune 500 use it in some corner of the org, this is the machinery that got it there.
Transcription is table stakes now. The real fight is over what happens after the recording stops.
Then there is Granola, which built a $1.5 billion company by refusing to do the one thing everyone else does. Granola does not send a bot into your meeting. There is no awkward third participant named "Chris's Notetaker" sitting in the corner of the Zoom grid. Instead the Mac app listens to your computer's audio locally and, crucially, enhances the notes you were already typing yourself. You stay the author. The AI just makes you sharper.
The founders are Chris Pedregal and Sam Stephenson, and they started building in London in March 2023. Pedregal is a repeat founder whose earlier companies, Socratic and Stack, both ended up inside Google. Granola launched publicly in May 2024 and then did something that does not usually happen: it became the app that venture capitalists and founders quietly recommend to each other, the kind of word-of-mouth that no marketing budget can buy.
Founder's notePedregal has said that "building in London in 2024 feels a little like building in NYC in 2014," a line that reads like a real estate tip and lands like a thesis. The money agreed. A $20 million Series A led by Spark Capital gave way to a Series B at a $250 million valuation in 2025, and then, in March 2026, a $125 million Series C at $1.5 billion. Zero to unicorn-and-a-half in roughly three years, on the back of one contrarian design decision.
Fathom took the oldest trick in software and executed it with unusual patience: give away almost everything, and trust that people will not want to leave. Its free tier offers individuals unlimited recordings and summaries, which sounds like a rounding error until you realize how few competitors dare to match it.
Behind Fathom is Richard White, who spent about a dozen years building UserVoice and, before that, co-founded a startup in Y Combinator's very first batch. Fathom itself went through Y Combinator in 2021 and became a rare direct investment from Zoom's own Apps Fund, which is a bit like a restaurant investing in the chef who set up a stall out front. The generosity paid off in a growth curve that reads like a dare: roughly $100,000 in ARR in 2022, then $1 million, then $10 million, then $30 million by 2025.
Nobody grows up wanting to take meeting notes. That universal reluctance is now a multi-billion-dollar market.
Read AI asks a question the others politely avoid: not just what was said, but how it landed. Founded in Seattle in 2021 by David Shim, who previously ran Foursquare and sold his earlier company Placed to Snapchat, Read AI layers engagement and sentiment analytics on top of the usual transcript. It scores participation, tracks attention, and tells you, gently or not, that the room checked out at minute 14.
Shim describes these assistants as "intelligent ghosts" sitting in on your meetings, which is either reassuring or unsettling depending on how your last one-on-one went. The market found it valuable enough to back a $50 million Series B at roughly a $450 million valuation. Read AI has also pushed past meetings into email and messaging with a search copilot, betting that the mood of your work is not confined to the calendar.
Gumdrop is the odd one out, and the most quietly clever. Every other tool here is obsessed with the aftermath, the recap you read once and forget. Gumdrop cares about the five minutes before. It sends you a briefing shortly before each meeting, pulling from your past conversations, emails, and contacts, sometimes even a photo of the person you are about to meet, so you walk in already oriented.
The founder is Paul English, who co-founded Kayak and then wrote a book about meetings, which qualifies as either commitment or obsession. With eight or more meetings a day, English built the thing he personally needed: a way to show up engaged instead of playing catch-up. The taglines, "Drop into every meeting ready" and "Meet with confidence," are honest about the promise. It is worth a clarifying note, because the names collide: Gumdrop the notetaker is not Gumloop, the separate no-code automation platform. Different companies, similar sound, easy mix-up.
In-house transcription at massive scale. Your company's searchable memory.
Fred joins every call and feeds your CRM. Built for sales teams.
No bot. Enhances your own notes locally on a Mac.
The most complete free tier in the category, by design.
Engagement and sentiment scoring across meetings and inbox.
Briefs you before the meeting, not just after it ends.
There is a tell that a category is growing up, and it is this: the products stop competing on the same feature. Three years ago these apps were all racing to prove they could transcribe a call accurately. That fight is over. Everyone can. So the six of them scattered into different bets, and the bets reveal something about how AI is actually landing in daily work.
Otter and Fireflies bet on memory at scale. Granola bet that the human in the room should stay the author. Fathom bet that the fastest way in is the cheapest one. Read AI bet that feelings are data. Gumdrop bet that the meeting is won or lost before it starts. None of them is wrong, which is why all of them are growing.
If you want to understand where AI is genuinely useful right now, skip the chatbots that promise everything and watch what quietly joins your next six calls instead. The revolution is not loud. It is ambient, patient, and already in the room, taking notes so you do not have to. The only real decision left is which of these six you trust to remember for you.