It is 9:00 a.m. and three calendars just collided. Somewhere in that pile-up, a small bot quietly clicks "join." It will listen to all three, write all three sets of notes, file all three lists of action items, and have an answer ready before lunch. Nobody thanks it. That is exactly how Fireflies.ai wants it.
Walk into any modern company today and you will find a strange new colleague on the invite list - one that never speaks, never interrupts, and remembers everything. Fireflies.ai is the AI meeting assistant that joins your Zoom, Google Meet and Microsoft Teams calls, records them, transcribes them in more than 100 languages, and hands back a tidy summary with action items attached. The pitch is almost insultingly simple: you show up, it does the paperwork.
By 2026 the company sits in rare territory for an AI startup - a billion-dollar valuation reached without the usual bonfire of venture cash, and a product used by teams across more than a million organizations. The firefly, it turns out, is a workhorse.
"The #1 AI assistant for your meetings - transcribe, summarize, search and analyze every conversation."
Meetings eat the day. Then they erase themselves.
Here is the quiet scandal of office life: we spend hours in meetings, and within a day most of what was said is gone. Decisions blur. Promises get misremembered. The person who took notes was also the person trying to talk, which is to say the notes were never very good. Companies have run on this kind of cheerful amnesia for a century.
Krish Ramineni and Sam Udotong looked at that and saw a different shape of problem. The issue was not that meetings happened. The issue was that meetings produced information no system ever captured. A conversation is data - speakers, decisions, commitments, sentiment - and it was being thrown away the instant the call ended. Their bet was that natural language processing could catch it before it vanished, and make it searchable forever.
Two thousand dollars each, and a willingness to fake it honestly.
Ramineni and Udotong met across the Penn-MIT corridor and started the company in 2016 with roughly $2,000 apiece. There was no AI yet capable of doing what they promised, so - in a detail the company now tells on itself - the early "AI" was partly the two founders, taking meeting notes by hand and emailing them back fast enough to look automated. It is the oldest trick in the do-things-that-don't-scale handbook, and they ran it for years.
"They each put in about $2,000 and bootstrapped for roughly four years - doing by hand what the software would later do by itself."
Most founders would bury that chapter. Fireflies frames it as the point. The product had to be good enough that someone would pay for it before the technology was elegant enough to deliver it. The hand-typed notes were a promise being kept on credit. When the models finally caught up, the company already knew exactly what customers wanted, because it had been delivering it manually, one painful transcript at a time.
The funding followed the product rather than the other way around. A seed round with Khosla Ventures and Canaan, then a $14M Series A in 2021 - about $19.1M in total. And then, notably, they mostly stopped raising.
The Fireflies Timeline
// from hand-typed notes to a billion-dollar bot
Krish Ramineni and Sam Udotong start Fireflies, taking meeting notes by hand while the AI catches up.
Khosla Ventures and Canaan back the bet on NLP-powered meeting capture.
Total funding reaches ~$19.1M - and then the company largely stops raising primary capital.
Fireflies turns profitable while still posting triple-digit year-over-year growth.
A secondary tender offer values the company at $1B; "Talk to Fireflies" adds real-time web search to live meetings.
The bot is on a million-plus companies' calendars - quietly, constantly, taking the notes nobody wanted to.
It joins. It listens. It hands you the meeting back, organized.
Strip away the branding and Fireflies is a loop. A bot joins the call. It records audio and video, identifies who is speaking, and produces a transcript. From that transcript it generates a summary, pulls out action items, and tracks who talked for how long and in what tone. Then it makes all of it searchable - which is where the loop becomes a habit.
"Let Fred review your meetings and come back with answers to any question you have."
The nickname matters more than it should. Internally the assistant is "Fred," and you summon it with a wave - "Hey Fireflies" - the way you would flag down a colleague. That is the whole design philosophy in two syllables: not a dashboard you visit, but a teammate you talk to.
The proofA unicorn that forgot to burn the cash.
Plenty of AI companies have reached a billion-dollar valuation. Fewer have done it while profitable, and fewer still while having declined to raise primary money for years. In June 2025 a secondary tender offer - not a fresh round - valued Fireflies at $1B, mostly so long-serving employees could cash out some equity. The company has reported profitability since 2023 and triple-digit annual growth on top of it.
Valuation: the unusual climb
// disclosed company valuation at key moments (USD)
The 2021 bar reflects the Series A round; the 2023 bar marks the profitability milestone (valuation undisclosed, shown for sequence); the 2025 bar is the reported $1B tender-offer valuation. Heights are illustrative.
The customer story is broad rather than glamorous. Fireflies does not lean on a handful of marquee logos so much as on sheer spread - teams across more than a million companies, sales reps and recruiters and support leads and finance analysts, plus a reported three-quarters of the Fortune 500 with at least some footprint. It is rated 4.8 out of 5 on G2. The pitch to a skeptical CIO is unromantic and effective: SOC 2, HIPAA, GDPR, admin controls, and the boring guarantees that let a big company say yes.
Headquarters
1355 Market Street, San Francisco, California
Founders
Krish Ramineni (CEO) & Sam Udotong (CTO)
Funding
~$19.1M total; $14M Series A (2021); Khosla, Canaan
Revenue
~$11M ARR (third-party estimate); profitable since 2023
The rivals
Otter.ai, Gong, Fathom, Grain, tl;dv, Avoma, Read.ai
Business model
Freemium B2B SaaS - Free, Pro, Business, Enterprise
When the note taker learned to look things up.
The 2025 milestone came paired with a feature. By partnering with Perplexity, Fireflies gave its assistant a live connection to the open web. "Talk to Fireflies" means that mid-meeting, someone can ask a question out loud and get a real-time, sourced answer without leaving the call or opening a new tab. It shipped to everyone - including free users - which is a confident way to give away your newest trick.
The missionDon't let knowledge die in a meeting.
The stated goal is plain enough: help teams capture, organize and act on the information inside every conversation, so knowledge is never lost to a meeting again. Said out loud it sounds modest. In practice it is a claim on something companies have always treated as unrecoverable - the institutional memory that used to live only in whoever happened to be paying attention.
"Reached a $1 billion valuation while staying profitable, without raising primary capital since 2021."
There is a useful irony in a company whose first AI was two tired humans now arguing that AI should do the listening. They earned the argument the hard way. They know exactly how much work a good meeting note is, because they did it by hand for years.
Why it matters tomorrowThe most valuable seat in the meeting might be the one that never talks.
If the bet is right, the transcript stops being the end product. It becomes raw material. A searchable record of every decision a company has ever made, queryable in plain language, feeding the CRM, the project tracker and the next quarter's strategy. The meeting stops being a black hole and becomes a database. That is a bigger idea than "good notes," and it is the one Fireflies is quietly building toward.
Back to that 9:00 a.m. calendar collision. Three meetings, one overwhelmed human, and a small bot that clicked "join." An hour later the human has three summaries, a list of who promised what, and an assistant standing by to answer questions about all of it. The meetings did not get shorter. But for the first time, none of them disappeared. The firefly kept the lights on.