Fireflies AI: How to Build an AI Unicorn
The best time to start an AI company? 6 years ago.
(Yes—before foundation models even existed.)
AI has come a long way. The progress in just the last year has been astounding. But building a truly great company takes more than riding the wave of AI agents and automation.
You need foresight about where the technology will be 3-4 years from now. You need deep context about your users. And you need to develop products that really solve the last mile of their problems, which is often very hairy and messy. It takes tremendous vision, technical proficiency … and extreme user empathy.
That’s why some of the most valuable AI companies today were started by founders who began working with this technology before it really worked. When the “Attention Is All You Need” paper dropped in 2018, only a few intrepid teams saw the long game i.e. what could be possible in 5-6 years not just at the current time.
They spent years adapting the tech to fit real customer needs. And in doing so, they built something far more powerful than just software—they built tremendous customer knowledge and awareness. Which is the context needed today to build truly great products and companies.
I was lucky to meet two of these visionaries early: Krish Ramineni and Sam Udotong, the co-founders of Fireflies.ai. Back then, I had just invested in Groq—another early bet on the future of AI, which no one had heard of either.
At the time, voice transcription simply did not work well. Many smart people were working on this very technical problem and trying to build new types of ASRs (Automated Speech Recognition engines). This took tens of millions in invested capital. Krish and Sam foresaw a world where not only did voice transcription work but that voice conversations would be one of the the biggest sources of valuable data on the planet. But they didn’t even stop there. They wanted to build workflows that derived insights from this voice data and automated some of the work that we do everyday. In 2019!
This was highly contrarian in a world where we were all still cursing at the IVRs on our phones when we called Comcast for support …
I jumped at the chance to be an Angel in their Seed round, and proudly led their Series A, 18 months later when I was a Partner at Khosla Ventures. Since then, they haven’t needed to raise a single dollar, they’re growing users and ARR like crazy, they are profitable, and still have a long runway ahead.
Looking back, their clarity of thinking was extraordinary.
Even more importantly, the team at Fireflies continues to innovate at the edge of what is possible in AI, but from a huge position of strength. They foresaw what would be the increasingly competitive battleground of “AI Notetakers” and starting building their next product iteration over 12 months ago. Today, they launched their next chapter - bringing AI teammates to life! Beginning today, all users will have access to Talk to Fireflies, a voice-activated experience that lets you ask questions and get answers about what you are talking about in your meetings - in real-time, in your meetings! It’s like having your own intelligent project manager / analyst / assistant on every call!
Congratulations to the team for continuing to push the envelope, eschew large capital raises and go beyond AI hype, to deliver production-ready, agentic AI that drives real outcomes. This is how you build category-defining companies… and AI decacorns!
If you’re building with AI today—not just for what’s hot now, but for where the world is going in 6 years—I want to hear from you.