Did you know that the models were gonna get this good when you started the company? Or how have you always kinda mapped the future of progress onto the roadmap that you have today and try and skate to where the puck is going?

There was no ChatGPT.

There was no GPT-1, GPT-2.

I believe OpenAI was started right around that time, and it was more of a research experiment.

One of the things that when we launched Fireflies was, hey, the notetaker back then was not gonna be good enough, and every time we go to an investor, they're gonna ask a thousand questions about, "How do you think that your model is gonna do better than a human at synthesizing data?" Like, summarization is a such a hard problem, and it cannot be solved.

This is twenty seventeen, twenty eighteen time period.

Uh, now it's obvious.

It's like, does Fireflies run on Fireflies? And I'd love to hear about, like, the internal agentic engineering you had to do to make that all work and how that manifested in the product that you actually built.

I mean, in the recent events, we've also seen AI being weaponized in defense departments, right? We had this conversation, I think more than a year ago as well, where we talked about what can India do from a vantage point? What can India do to get its AI strategy right? Has your opinion changed?

So much has changed in the last 15 months, and I'm a firm believer now more than ever that those that leverage AI are going to come out ahead.

And what I mean by that is in the next decade, AI is going to influence not just who are going to be the best companies, it's going to define who are the countries that are going to come out ahead.

That doesn't mean that every country has to follow the U.S. model or China model.

One thing that I've always believed in is India has incredible talent.

But the capital investment that you're seeing in a few years ago, it was $280 billion worth of capital just went into AI investment in the U.S., whereas in India at the same time was around $4 billion.

If you look at the developers on GitHub, you had about 20 million developers in India and the US was around 27, 28.

But still for a common user, why does it not feel like singularity yet? Tell us what is your understanding?

Has your opinion changed?

So much has changed in the last 15 months, and I'm a firm believer now more than ever that those that leverage AI are going to come out ahead.

And what I mean by that is in the next decade, it's, AI is going to influence not just who are gonna be the best companies, it's gonna define who are the countries that are gonna come out ahead.

That doesn't mean that every country has to follow the US model or China model.

One thing that I've always believed in is India has incredible talent, uh, but the capital investment that you're seeing, uh, in a few years ago it was $280 billion worth of capital just went into AI investment in the US, whereas in India at the same time was around $4 billion.

If you look at the developers on GitHub, uh, you had about 20 million developers, uh, in India, and the US was around 27, 28.

But by April of this year, India has already caught up and has 27 million-plus developers.

Strawberry

So tell us about the start of the journey.

For us, fundraising was actually a very difficult process in the early days.

The smallest check we could get at the time, or any check for that matter, was a $25,000 check from a small angel accelerator.

And we used that money and had it last for over a year in San Francisco, one of the most expensive cities.

It's not that we didn't want to raise money in the beginning.

We just couldn't.

And then after a certain point, we said, maybe we're not built for raising venture capital because everything I would hear is about people overpromising, founders saying they're going to do this and that.

Can you give some examples of some of the things that you actually did to try and keep your costs down?

How does the approach to something like an AI note taker now with the, you know, the models that are available differ, both in terms of, like, what is easier, but maybe also there's certain things that are challenging now which you didn't have to think about before because of how the technology has- has, uh, evolved? Um, would- would love to hear that side as well.

If you're using transcription engines today like Whisper or one of the ASR providers today, and even some of the big ASR providers from the Googles of the world, Microsofts of the world, you don't have to massage the output as much.

That's ex- exactly the word that we had to think about, 'cause the raw output was pretty crappy and we had to fix the grammar, we had to fix the punctuation, uh, we had to figure out speaker identification at the time.

So you're just getting a, you know, garbled up string of text and you're trying to make sense of that.

And, uh, you also have to figure out how to identify certain type of words, because for certain industries, certain work domains, there are different types of words that are out there.

How do you deal with filler words, pauses, ums? Like, we speak a lot of filler words.

I feel like working on Fireflies has improved the way I speak and helped reduce the number of filler words I- I use because the first versions of Fireflies, every time I would read the ums and the uhs from my own meeting notes and recordings and transcripts, I'd be like, "Wow, I said it that many times?" So we ended up building a little filter back in the day that told me, like, how many filler words I used in a meeting.

Uh, you know, what, what has enabled that at this, at this point in time?

How have you seen that and shaped that? And how has that helped you with your growth?

When I look at my own behavior, I realize how dependent I am on AI every single day, whether it's Chad GPT, whether it's Fireflies or a dozen other tools.

I actually make it a point to try at least two to three new AI tools every week.

And maybe not directly in our category, but other categories.

I'm absolutely amazed by some of these video and image generation tools like Higgs Field and Nano Banana.

Yep.

It's just so many great tools out there.

It's

right there's basically like as a human using like very simple like getting to the heart of what the problem is and able to demonstrate people the symmetry to people that the solution is a lot easier than they think but uh before we step into that tick-tock rabbit hole right we got krish our first guest on bottleneck and we want to hear a little bit more about yourself what have you been working on what are some of the bottlenecks in your industry

Well, we've been working for almost nine years now on building an AI assistant for people that can follow them around to their meetings, help write down notes for them, remember what to do, and be really, really good at helping you have perfect memory.

That was the problem we wanted to solve.

Now, we've had so many technological bottlenecks from the day we started.

On day one, people would say, this technology is not good enough, right? Transcription, when we started, was only 80% as good as what it is today.

So at the time, we had to change the framing.

It's not about having perfect transcription or perfect nodes.

What are the cognitive constraints which this can help resolve?