Signal Dialogues #02 | Wispr Flow CEO Tanay Kothari on Beating Apple & Google at Voice

So you started Whisper 4.5 years back. Little bit about the early days of the journey.

When I first saw the Iron Man movie, I wanted to build Jarvis. And so me and a friend of mine built what was then one of the world's first voice assistants. 2.5 million users, 6 months, and then Google actually shut us down because they didn't like what we were doing.

Feels like one of your earliest insights was that these organizations are just set up wrong.

Organizations are set up wrong. Whenever somebody asks me like, why can't Apple build this? I mean, like, of course they can from a resources standpoint, but you need to build a different organization for that.

In consumer AI particularly, How much does the quality of these first 1,000 users matter?

Actually, don't think it matters in consumer at all, because in consumer, if we try to build a mass market product, I fundamentally believe that's how you die. But my key goal is with every type of customer that we build the product for, they need to feel that this product is for people like me. How do we make this human-computer interaction be as seamless as possible? So using your devices feels like talking to a close friend.

Hi everyone, welcome again to Signal Dialogues. I'm back. This is Akrit Vaish. I'm the founder of Activate, India's native AI venture capital firm. We work with young technical founders way before they're ready to start a company and invest right at inception in founders from India building AI companies for India. And for the world. So when we talk about young, technical, from India, for the world, I think I couldn't have had a more apt guest that fits all of those words in Tanay, who is the co-founder and CEO of Whisperflow, which today feels like one of the most loved consumer AI products out there. You know, I think, Tanay, we've obviously been chatting this whole week about your journey, and I feel there's so much to unpack that it just best comes from you. So why don't we start a little bit about if you had to introduce Tanay and Whisper in 60 seconds.

That's a lot. So, I grew up in Delhi for most of my life and been working on Whisperflow for the last 2 years, but generally been in this space for the last 18. Which is what does the next generation of personal computing look like? Because growing up, when I first saw the Iron Man movie, I wanted to build Jarvis. And so me and a friend of mine built what was then one of the world's first voice assistants. This is before Siri, before Alexa. To people, it just felt like magic. 2.5 million users, 6 months, and then Google actually shut us down because they didn't like what we were doing. But fast forward many years, builder at heart, built a number of mobile apps, web apps. And the key thing that drove all of that was whenever I saw anybody on their phone or their computer, it just felt so effortful and so mechanical and not how I want technology to feel like at all. And so with Whisper, the key goal that we have is how do we make this human-computer interaction be as seamless as possible? So using your devices feels like talking to a close friend. Amazing.

You did nothing better than 60. I mean, Tanay Kothari, like it's been so awesome getting to know Whisper, the product, and you. And you know, one of the things that strikes me every single time, and most of the audience that watches this is either founders or people who want to be founders in AI. One of the things that consistently strikes me about you is the depth, right? Like we always talk about the depth of understanding around this product, both technology, product, your consumer. So let's start a little bit at that. How did voice happen to you and when did you, like in terms of technology? And then what did it take to become what I would like to think is today 99th percentile plus in this area?

So voice, interestingly, was one thing I had a very strong belief in since 2008.

2008?

2008.

Tanay Kothari, you were what, 18 that time? No.

10.

Okay, 10. I was close to 18. Sorry, I forgot.

And so the—

so you were 10 and in Delhi?

I was 10 in Delhi. I was in 5th grade. And it really struck me where I was like, what I want, all I wanted was for my computer to be just like a smart friend who I could talk to and we could just do things together. And so that was the first inspiration for the first product we built. That was VoiceOnly. And the key idea behind voice stuck with me throughout, basically now it's been 18 years. But there was a big realization that I had about 4 or 5 years ago when I was thinking about the next phase of technology, which is— so the world we live in now, everybody's glued to their phones, right? And we all talk about, oh, we want to spend less time on our screens and so on. But the best solutions we have is screen time apps, and they're a band-aid to fix this problem that's otherwise not getting solved. To me, the best way you move away from display-first technologies like your phones and laptop is a voice-first technology. And so the kind of world I want to build is where you go downstairs, you're walking on the street, and you see everybody just looking up. Nobody's hunched over like this anymore. And so that is the key vision that we started Whisper with 4 and a half years ago. So something is—

let's just go back to 2008 for a second, right? Yeah. In all seriousness, you were obviously in 5th grade and I was just about finishing college. Products existed.

Products existed.

So you had Siri, which was actually—

Siri actually came a few years after.

2008 is iPhone launch.

2000, uh, iPhone launch happened, I think a year before.

2008, and then 2009 was, I think, the App Store launch, something around that. So Siri was maybe 2010, 2011.

Siri was 2011. Yeah.

So what was the product that existed back then?

So the only thing that existed back then was Google's speech-to-text.

On that thing where you press the mic button and—

when you press the mic button and you speak, and that was only on Android phones, and that only some Android phones had that.

So just doubling down a little bit on voice, and my observation, because I also come from a similar kind of trajectory where I started Haptik thinking about similar themes, why do you think For the better part of 8, let's say 10, 10 to 23 or 24, when this started becoming popular, for 13, 14 years, all these big companies with so much capital, so many resources, the best talent, just couldn't get this right. And even till today, I'll tell you an incident just this right now on the way over here to the office this morning. You know, I use, we all use Whisper a lot. And then, you know, because of muscle memory, what happened is that while trying to press the Whisper button, instead I pressed the mic button on the keyboard. Yeah.

Which was the Apple dictation.

Which is in my case, Android. Okay.

Android dictation. Yeah.

And I pressed that mic button and then I started speaking and I could instantly notice such a big difference.

Yeah.

Because you're used to using Whisper and I'm not saying like any specifically any other tool, but 14 years, give and take, large companies, people, and still continue trying it. Just from a technology perspective, like what do you think has been the unlock for industry and more specifically for Whisper?

So this is the question I had, right? Because Apple, Google, all of these large companies have half a trillion dollars in their bank. They have some of the best researchers, like tens of thousands of them. And I had initially started building Whisper's hardware because I thought everybody's of course going to solve this problem, how could they not? The key thing I realized when I did talk to them and I did talk to the heads of these voice teams inside these orgs was— it's a very interesting realization that the way that structured these companies was there was a research team that built these models and there was an engineering team that deployed these models. And there was a product team that talked to the customers. And so the people who are actually building these models never talk to the users. So when I talked to the researchers, they said, oh, this problem is solved, works great, we've solved all the issues, we have such good word error rates. And these conversations are happening even in 2024.

Even today, I'm guessing.

Even today, there was an exec from one of these companies going around and they're saying like, oh, we've solved voice, this is great, works really well. That caps. I think they did the best that they could with the information they had, but this is a fundamental thing where, because your research team is just working in this black box with your benchmarks and all of them that you've set up, which are so different from how the average person uses it. And so an average person is not going to say, hi, comma, Akrit, how are you? Question. Like, that's not how you say, hey Akrit, how's it, um, um, how are you doing? That's how they talk. You ramble, you change your mind.

It's so fascinating, man. So your insight really was, of course, technology, but your— feels like one of your earliest insights was that these organizations are just set up wrong.

Organizations are set up wrong. Wow. And so whenever somebody asks me like, why can't Apple build this? I mean, like, of course they can from a resources standpoint, But you need to build a different organization for that.

So your— the other day also we mentioned this, you know, for the lack of better words, I hate this term, but let's just use it for simplicity. Your moat.

Yeah.

From the outside, people may think is, oh, they have some great model that they've built and they have some insane technology, which holds true, but sounds like your moat has simply been just better structure and execution.

For this specific problem, yes. And I think that's one of the reasons why I think startups can win against the large players is because when you are just deeply obsessed with solving a specific problem, you build your whole company around to be the best person to be able to do that. And so for example, at Whisper, our researchers are talking to the users, they're shipping new things every single week. And so that research engineering product is just one person inside our company and every single researcher is like that, which is a very different way to build an ML research org than any other company has built before. And the thing is for a product—

I mean, retrospectively, this sounds so trivial, right?

Yeah.

But like that, and that to your point is the beauty of startups that you land into something that in hindsight feels like Why does it— why do more people not do that? Yeah, that's amazing. Okay, so fast forwarding a little bit, so you started Whisper 4 and a half years back?

4 and a half years back, yeah, August of 2021.

Okay, awesome. And a little bit about the early days of the journey.

So this time, I have like August 2021, I'll paint a picture. GPT-3 had come out. ChatGPT did not exist.

ChatGPT is in 2023, right?

So we're before ChatGPT. And the key insight that I had is computers can finally understand natural language when I saw the first LLMs.

GPT-3.

GPT-3. And we said, okay, everybody's going to be talking to their computers in the next few years. What are the biggest problems that nobody else is working on that we need to solve? So I thought, okay, you need to build really good voice input. Of course OpenAI is going to build that. Then you need a really good voice assistant. Of course Siri and Alexa are going to improve. Like, this is my 2021 mind. So the thing that you will need is you want to be able to use voice when you're around other people so that you have privacy, you don't disturb them. And so we decided to solve the problem of silent speech. This, over the next 3 years, got us to build a team of about 40 people, PhDs in neuroscience, signal processing, electrical, mechanical, machine learning. And we actually built the world's first thought-to-text device that was non-invasive. So imagine Neuralink, but you don't need to get brain surgery. It's just a Bluetooth earpiece that you wear. And we had the best people in the world. There were professors from Ivy League schools who gave up their tenureships to join Visper full-time. Our chief scientist used to manage 5 Nobel Prize winners at Bell Labs. These are the people who we had brought together to build this. And when it worked in June of 2024, we connected with ChatGPT, Siri, Alexa, and they all sucked.

So let's talk, uh, so how much capital had you raised by then for this?

We had raised a $21 million seed.

Which was for this thought-to-text?

Which was for this thought-to-text hardware device, right? So Whisper at this time is a deep tech brain-computer interface hardware company.

And what was it called?

It was called Whisper.

The product was called Whisper.

The, yeah, so we had two SKUs. They were Whisper Pros.

And the name makes sense now because the silence.

Yes. So that's where the name came from. And so there were the Whisper Pros, which were headphones, and Whisper Airs that were smaller earpieces.

Got it. So what happened when you started connecting this with—

so we started testing all these devices and we realized that they all sucked and we needed a way to go from your rambles into something that is structured. And ready to send. And so we built an operating system for this called Flow. Oh wow. And that we started using, and then I had to send it to a couple of friends to test it out. So that's when I decided to package it in a desktop app, and I sent it to them, and I was like, you know what, instead of silent speech, let's also make it work with voice. So it initially did not use the microphone at all. It just used our headphones. I added voice as an additional thing to test it out. And these guys loved it. These 2 friends send it to 5, send it to 10, and suddenly I'm looking, we have 100 users from out of nowhere. And the biggest thing I realized, like, remember the problem I was telling you that we wanted to solve? How do you use it around other people? So all 40 of us used to be in an open office, right? All sitting together. And one day I just walked in and the thing I saw is everybody had a mic on their desk, like these United Nations kind of mics coming up to your mouth. And they're just like silently speaking to this, to their computers, and everybody in the office is doing that. And that just reality slapped me in the face where it just brought me back to the problem that we were looking to solve. And I was like, wow, this solves it. Wow. Everybody in our office is using this and our hardware devices put on his side.

Just picture that moment right now. Everybody's like got these like those cord mics hanging up to your—

Yes, exactly. And so I saw that and that's when me and Sahaj took this hard decision to pivot the company completely. And so overnight we went from 40 people to 4 people. We were no longer a brain-computer interface hardware company. We became a consumer software company. And then we rebuilt everything from scratch. And at this point, all we had was the software product.

And this is '24 summer. So only 2 years.

This is 2024 August.

18 months, a year and a half.

18 months ago, we decided to make the pivot. Feels like ages ago now when I think back. And so we did that. At this time we had 100 daily active users for this product.

Basically your friends and who they sent it to.

Basically these friends. And then Sahaj and I had a conviction that we saw both of our behaviors change. And that's one thing I truly believe about consumer software is the best consumer softwares create behavior change in a way where as a person you can never go back. Basically transforms you to a new world. Doing that is also both incredibly hard, but also if you can do that, it builds an incredibly strong business. So we told our investors, we actually had a lot of money in the bank, half the money was still there. So we did the pivot not because of money or investors or anything else, it was just he and I believed that we needed to solve the voice problem first. And so we told them, hey, we're pivoting to this. Now this time, remember, 2024. Everybody thought we were out of our minds because we were building a dictation product. And they're like, dictation exists on your phone and laptop, it comes from free. Mac literally has a microphone button. Like, what are you building?

I can't even imagine, man. Like, yeah, exactly. Like, if you were to tell me at that point, and this is exactly not that far, I'm trying to think of my worldview at that point. At N24, I'm investing very actively, at least in India. I'm seeing a lot of stuff in the US. I'm there in the Valley every 3 months. If you told me that, hey, listen, there's this guy trying to build a dictation app. We'll be like, okay, anything else?

Yeah, no, exactly. That was everybody's reaction. And so our investors are fantastic. They supported us. I don't think they had a lot of belief in what we were building. I was like, you know, fine.

Investors have to be like that, especially once they're more— if an investor or a VC tends to be a founder backer, then you have to just play along, right? You have to say, okay, I believe you. And then you're only going to leave the office and be like, We'll see what happens.

I know they were supportive like throughout. And so that was one thing that I really appreciated. Love all of my investors. And so we started building it. And then the thing I told the team is gun to your head, we're going to launch this product in 8 weeks, right? 8 weeks from a demo prototype we'd built in-house, no marketing team, no brand, no nothing. And then we basically did that, built a whole marketing function in under 8 weeks, planned for a launch, had a full launch video. I actually got my brother up. He's, he does video editing. And so he and I just shot the launch video over a weekend and that went viral. We had millions of views on Twitter. I think the first launch video hit about 4 or 5 million views. Massive launch on LinkedIn, and our numbers started skyrocketing. So this was in October of 2024 when we first launched Whisperflow to the public. Since then, it's been kind of just this incredible rocketship journey where we basically had to create a new category. We had to tell people why this was something that they should use. And now it's a lot more of a, you know, norm. And somebody says Whisperflow, a lot of people have heard about it. There's a very strong brand there. There. But back then it was really hard for us to acquire new users because everybody was like, oh yeah, I tried Siri and it sucked. I don't want to try voice again. So voice was this, voice was this field that just had a really bad reputation because consumers had just been let down for the last 20 years.

Man, that's crazy. So let's then talk a little bit about post, uh, and then you called it Whisperflow because Whisperflow, we combined the names. It just came naturally. Let's go back to sort of launch plus 3 months. One of the questions that often, I'm sure as a founder you've been thinking, you were thinking about this back then, we all think about, right? That, okay, look, that first inflection point. Yeah. How did that happen? Like, was it genuinely just, how do I explain it the best way? Was it manufactured luck to a great degree? Did you do anything in those first few months that, you know, you think was some of those smaller things that make the big— that made big differences?

A bunch of them. So the first thing that I very strongly believe in is at the end of the day, you're giving a product to humans and humans are very emotional creatures. So if you ever are sitting in a kind of meeting with me at the company and I'm doing a PRD review or anything like that, it all comes back to TK, how does this thing make the person feel in the specific funnel? And so what that also looked like is for the first 500 users that we got, you actually couldn't download the product. You set up a call with me. I was on there with you for 20 minutes. You would screen share and go through the entire onboarding process. And the thing I was looking for—

you still do that to some degree?

I do that to some degree. Like what, yesterday night somebody was installing the app at the dinner we were at and I just stand over their shoulder and I just look at them and then I found a couple of things to fix in the onboarding. Then I messaged my team right after. And then I do that at airport check-in lines or wherever I get people to download the app because I just want to see, you know, because like what is non-obvious to people, how are they feeling? But the main thing is when I was looking at these people onboard onto the product, the key thing I'm looking at is their expression. Where are they confused? Where is there a smile on their face? Where are they feeling like, you know, a little impatient to go ahead? And we did so many improvements on their onboarding side. Because I wanted that to be the first impression. You know, it's like somebody walks into your house, the first thing they see, they set an impression. When they first meet you, first 5 seconds sets an impression. For me, that was the product. Like the first onboarding has to set an impression. And the key thing I realized is if you can make the person feel extremely excited and just feel like they have unlocked a superpower, then You can do so much more with them. And so my full focus was the onboarding funnel we worked on for 3 months. The next thing was figuring out how to reach these early people. And so set up with this hypothesis ki it seems like there's a lot of founders right now that are using this product, so let's target them. And so my key thing was I would go to them and be like, ki diya hai, what are the couple of recent tools you've tried out? They would tell me. And then I would ask the most important question, which is, where did you find them? Because that would tell me a validated channel that they trust to discover new tools. It was subreddits and it was newsletters. So that was the first set of things that we started off with.

Not so much Twitter.

Not so much Twitter then for this like first— yeah, yeah, yeah. The next batch of people was Twitter and LinkedIn, so I scaled over there. But that's how we got to our first 1,000 users before the launch. Because my main thing— and these people I onboarded because I wanted to make the product perfect before the launch. Perfect. And so that's how the first couple of months looked like ahead of the launch.

Tangential question, but a very important one that I think about a lot. Do you think that in consumer— I think I know the answer, but I'll tell you where I'm going with this. Do you, in consumer and consumer AI particularly, How much does the quality of these first 1,000 users matter? So where I'm going with this is that it's a very common thing in B2B, right? B2B software, AI, Gen AI, that, you know, choose your design partners, design partners very carefully, work with them hand in glove. And that's very important because they'll define how to build the product. Did you have a similar lens on when you were rolling this out?

I actually don't think it matters in consumer at all, because in consumer, I think that humans are fundamentally very similar. And this is, again, you're thinking about it from the human psychology layer, from like that emotional standpoint.

But like, just to challenge that a little bit, right? Let's say if you were not in Silicon Valley and you did not go to founders and you went to say, you know, a student in India, yeah, I'm sure the feedback and the way they would do things would be very different.

It would be different. And we'd start off with the students bucket. We would start off with a completely different ICP, but that would be fine because we'd learn from them and we decide how we wanted to scale the business because we started with that founder ICP.

Did you also start with the founder? So did you start with the founder ICP? There could be two reasons. One, the obvious one that this has, this ICP seems like has the largest problem that we can solve the best, which is the first principles way of thinking. Founders are super busy typing all the time, so let's solve it. Was there an angle of also this could be a very influential consumer base to drive further growth?

I'd love to say I had that hindsight. No, founders we landed upon because of convenience first. But then, so after we did the launch, we actually saw a very surprising thing, which is our users came from all walks of life. We had lawyers, Doctors, students, we had designers, we had real estate agents who were using Whisper every single day. And basically when we talked to all of them—

isn't like a large segment today just coders?

Coders make up for about 12%. Yeah.

Or was it you just have this on and you're like talking and just coders talk about it a lot.

So that's why probably I think if somebody had to guess, they would be like, oh, 80% of Whisper users are coders, but like, no, it's 12%. 5% are lawyers. So there's 2.5— like, there's only 2.5 times more coders than lawyers who use Whisper. And so then we actually defined 26 different ICPs that we could go and build the product for.

This is pre-launch?

This is right after launch.

So founder ICP was clear.

Founder ICP was clear. Then we found these other ones. And so by January, we had a list of them, and then we're sitting and we're like, okay, what do we do now? 'Cause if we try to build a mass market product, I fundamentally believe that's how you die. If we try to tackle all 26 ICPs at once, we will also die because remember the team at this point is very small. We're 7 people, 1 person in marketing. And so we decided to pick 3 ICPs, developers, creators, and founders. Founders expanded a bit into business owners, and here's why. One, these people are generally more tech savvy and often download more products, so it's great, it's going to be easier to get to them. But the second and more important point was they had a high K-factor, because if you get one of them, they will tell a lot more people, they'll tell their companies and so on. And they are those influential hubs where you can get more spread from. So that's how we picked these segments and that worked really well because what we saw happening was on Twitter and LinkedIn every day there were 5 posts from these influential people talking about how much they loved Whisperflow and creators were making videos just showing Whisperflow in every single one of these. They might not even talk about it, they were just using it and then every second comment was, what is this speech-to-text tool? So that was the second big wave we saw from focusing on those. And now of course we've expanded the ICPs that we target a bit more because we can focus on more people. But my key goal is with every type of customer that we build the product for, they need to feel that this product is for people like me. And that drives a lot of what happens on both from the product to the marketing and messaging side.

This is incredible. And this, I can imagine, you know, even Like I'm learning so much even after having known quite a bit about the company. Tell me a little bit about what does the average, not average, because there's nothing average these days, but what does your typical or day-to-day at Whisper look like in terms of how you organize your life, right? So like the metrics you track, what do you do first thing? How does the team work around you and the pace you operate at? I think like some of those things, because it feels like I consistently keep hearing this back again and again through this today and even the last few days that a lot of what is making Whisper tick today. So, you know, I had long, long time back seen this image, which was this arrow, like a big arrow. And it said the base and then it had multiple sort of milestones along the arrow. And then there was a tip. And if I had to take the Whisper sort of analogy, the tip of the arrow is what we see, right? As consumers, like, oh my God, what a lovely product. But there's so much at the foundation and the input and the execution that is going in as a company to make it work. So give us a little bit of a peek into that, that how day-to-day, what does it look like, what makes it really work?

Yeah. So I'll give you how I think about structuring it, because my every week looks completely different than the week before. So every Friday I have one hour that I spend thinking about what are the most important problems in the company that I need to go and solve. I can do today that will unlock the next level of growth, help us build the next mode. And I come up with 3 key priorities. 2 weeks ago, for example, when I was doing it, that looked like we need to scale up our B2B motion, we needed to uplevel the engineering team, and we needed to prepare more for the India launch. These are 3 things. My next week, every day starts off with 30 minutes of just reflection on the previous day and the key goals for that day that I have. This is actually insanely important for me to set up because my default is I'm a guy who I'm like, I can do everything and anything, right? But the problem that comes is I then plan for more things to happen in the day than I have time for. and I just keep working through the night not being satisfied with what I've done. So this thing helps maintain sanity because I pick these 3 things that I need to solve today, and then by the end of the day when I solve those 3 things, I can actually rest in peace. And that is huge to come back tomorrow fresh and be able to be my best self. So then I pick these things, and during the day I can spend this time anywhere from sitting individually with my employees and helping them in some ways. It could be a 4-hour block that I will set up sometimes just to deeply think about a new thing that we're doing and figure out how we want to approach it. Some days it's just meeting a lot of people outside the company to either bring them on as partners, potential hires, and so on. The way I've structured my day basically comes up from how I think about managing my energy and how I think I do my best work, which is I've realized for myself, I can be in one mindset in a given day. If I switch mindsets, things don't work. So if I'm deep in engineering and the product, I don't talk to anybody outside the company. When I'm talking to people outside the company, that's all I want to do for that day. Because I'm in just this zone of selling, convincing, all of that, that I need to make happen. I realized when I used to context switch those two, it would just kill my energy and I would be a potato by the end of the day, which is not what I want to be. And then the key thing I realized is this one thing that I started doing a lot this year. If I look at a certain team that's not doing well, a couple of years ago, I used to think ki, Oh, maybe they're not working hard enough, or maybe like the people aren't right. But now the two things that I come to is, one, I might not have set the right expectations with them. Two, I really wanna understand why this is not working as fast as I want it to, because either my expectations are wrong or they're not working as I think of. And so what I started doing, I saw one of the, the teams not doing well. And I just sat with them and we, I saw them how they worked with, for example, Claude Go to build new systems. And I realized a few very key points of inefficiency going on. So I worked with them, I told them, ki, this is how I would do this and so on. And so basically the feature that they took 1 hour to build, they could now do in 20 minutes. And the next week onwards, oh my God, the velocity that team operated at. And so a lot of times, I used to previously think like, okay, I need to hire more engineers because this team is unblocked and going faster. But now it's unlocked too, because I know I hired great people. I think I can work on them to make them faster. And that is a big thing that's led to a huge unlock because Whisper is still a small company, just 50 people.

Yeah, yeah. Amazing, man. You know, on this last point, so sorry, just recapping. I really like that idea of, okay, look, step back every week and look at sort of what your top 3 priorities look like. You know, how I think about just caveating that for founders is that per week, every day, or every quarter, or every month, that frequency needs to be determined for your business. Yeah. Right? Like in a lot of, like in an enterprise sales business, doing it every week is actually more—

You might probably come up with the same set of things and—

It's actually more harmful, right? Because you can, you have to see things play out. That's true. Like you have to see long sales cycles play out for some time before you come back and sort of say, okay, listen, we're going to go all into this as an example versus for you, it's a more frequent business. But on that last point, even at Activate as a venture firm, it's kind of been hilarious, right? All of Activate today and people see us doing so many things across the board. Like this conversation is not even directly related to my business. You know, you're different stage, different sector, different country, company, but even the entire firm besides the two partners and founders, Pratyush and me, is one full-time person.

That's incredible. Incredible.

And every few weeks, like why I found that analogy interesting is every few weeks, you know, I will just wake up being like, oh my God, this is so exhausting.

Yeah.

We need to hire and we'll go put out JD out, right?

Yeah.

Within 2 weeks, by the time we meet candidates and we make some progress, we've already done so much on Claude Code that now we don't need that person. And this has been happening consistently now for 6 months, right? If you just go in and spend some time and you've identified something and you automate it, the, just the need to have teams and people do mundane work is just, you know, not needed anymore. And same thing, like the last piece that you mentioned, right? That if you just sit with somebody, that 60 minutes can become 20 minutes. To ship a feature and then suddenly realize that, hey, listen, we don't need 10 more engineers, maybe we need 2.

Yeah, exactly.

Awesome. On that note, you said something that I picked up on. Tell me the— you can choose the number, 3, 4, 5, 2 metrics that you obsess about.

The top line metric that we track is what we call DAO current. So specific metric that comes from this underlying Markov growth model that a lot of great consumer companies like Duolingo use. I can go back into that and why that's helpful, but it basically is an incredibly useful way to model out your growth by merging in retention, activation, everything into a single model. But this is basically our proxy for an engaged user. So I don't care about the—

So DAU current.

DAU current.

And it's a Markov model for people to look up.

Exactly. And so Duolingo has this fantastic article on how they model their different users, and I'll tell you why this is the key metric I look at. My top line goal is not get as many people as possible to try Vesper, it is to get as many people as possible to build the habit of voice. And so if I get a user that tries this once—

we ran into one guy last night who's 100,000+ words, right?

Yeah, exactly. Those are the kinds of users I want to create. And so I'll tell you one funny thing. 3 months ago, we used to track ARR as a metric. ARR is not a top-line metric anymore. Wow. Because when that was a top-line metric and I would talk about that in every single all-hands, the team was like, okay, now we need to find ways to improve ARR. Then they started doing more monetization experiments. And I was like, guys, okay, we've limited resources. I would rather have you spend those resources in experiments to get people to build the habit of voice, figure out what's stopping them, make them more comfortable about it than trying to get them to pay us. Because if we get somebody engaged and loving the product, we'll get them to pay. And so there's not saying that the ad is not doing well. We're growing revenue 40% month over month, but actually we don't even talk about that because the things I talk about and the things that you track is the exact way your company operates. So this metric, DAO current, is users who use the product today, who have used the product one more time this week, and they're not a new user. So it's like a small segment of this, but what this does is this is not gameable because I can't spend $1 million in marketing, acquire a bunch of new users who drop off because that's not going to move my—

So the user who used the product today, he was he or she could have been a new user yesterday?

We keep a new user for 1 week. So for the first week you don't get to enter this category, so we have to retain the user for at least a week. And then these people, it's not like you're using it once a week, you use it at least twice a week and per day. And so because we add so many filters, that makes a very solid metric. And that to me is the best metric that is both very good to model and it is very good for tracking the key thing that we care about. And the key reason why we chose this metric and why modeling is really important is because whenever we launch any experiment, right, and so we have a very deep culture of growth experiments within the company, a lot of times you need to let an experiment play out for a while, right, 1 week, 2 weeks, 3 weeks. And yes, that's great, but there also it really lengthens your learning cycle. So what we build underneath every single one is a predictive model that is able to predict, okay, based on these early—

Growth and Dow Current.

Growth and Dow Current is able to predict how this is going to affect the key metrics that we care about. And that is incredibly helpful because that gets our marketing team very early numbers to improve performance marketing, that gives our growth experiments very good numbers to be able to scale this very quickly. So now I go into like very minute details, but this now fundamentally drives everything else. And of the 100 metrics or so we could have chosen to be our number one, this is the primary one we chose. And this is the one we start every single all hands with every week. This is where we are, this is how much we improved, and these are the things that we did to improve this number.

Amazing. Anything else? What's second level, third level?

What's the— Second level that's been top of mind for me right now, and this is a right now thing because I'm focused on building our B2B business is the number of companies.

You know, I want to talk about a reflection right now. We've been spending time, like I said, all week, right? And I want to kind of bring a few things together. You said that every Friday, right, you think about the 3 priorities. And I just want to reflect for you is that I can clearly tell that you've thought about this B2B thing recently. Because through this week you've mentioned this so many times, and even in just casual passing conversations, right? Not like formal, obviously, chats and all that, but even when offline, people have been hanging out, I've noticed this come up so many times. So just as a takeaway, I think for all of us, you know, one thing I can sort of imbibe, for the lack of better words, is when you're obsessed about those 2 or 3 things, it's just like ingrained, right? Like you're talking about it, you're doing this all the time, you're telling people, you're telling your company probably every like the company becomes that for those few.

I do get very deeply obsessed whenever I get into something and then I only talk about that because that is all I'm thinking about. Those are all the people I'm reaching out to, to just learn a lot more. That's, by the way, one of the things that I personally love about building.

Is that how you think about culture for everybody else in the company as well?

I think about culture for everybody else as— so I'll tell you the slight difference. So the one thing is, if you do something, just do it incredibly well to the best that you possibly can do. And so that is something I expect from every single person in the company.

I think depth is so underrated, right? Like in this world of like, uh, in the world of like 60-second reels where everybody is just obsessed with consuming content, it feels like, it almost feels like anybody who can become like 90th percentile plus deep in a subject. Just by design will have a higher chance of success.

Yeah, I think that's, that's huge. And so that's one thing I focus on is that whenever we hire any new person, the one thing I tell them in my first one-on-one is I just want you to be the best person in the world in the specific field that you're in. You tell me what you need from me to do this. If you want anybody in the world to come and be your advisor, be your mentor, I will go and find that person for you. But I want you to put in the effort to be able to build yourself to that point. And so there's actually a bunch of things that I personally do for each of these people to get them the right person. For example, we had somebody in our team who was going from becoming an engineer to an engineering manager, and I put him in touch with one of our angel investors who was the CTO of Dropbox to help him. Like, how do you do that? Because Dropbox had an incredible engineering— Yeah. And so that is something I do for every single person in the team. Now they have to come up to me and request to—

by the way, that's a side note, that's a great way to use angel investors.

Yeah, right.

Like, often enough, like, founders have this, uh, you know, call it this celebrity list of here's my VC, here's my angel. You look down the cap table and then, you know, they struggle to actually make use of most of them because you usually reach out, hey, listen, I'm stuck, I want an intro. But actually getting your team to potentially work with an angel actually is a great idea.

Yeah. So most of the angels we get, they were all ex-operators. And my key tell for every angel is, what is something that you can teach me? And that is what I look for in my first conversation. Is everything else apart, what are all things that I learned from this conversation and what was the rate of learnings per unit time? Because somebody scores very high on that, then I don't care, I want them as an angel on our cap table.

Awesome, man. We are right about to wrap up. So let's talk quickly, maybe next few minutes. About India and India, two perspectives.

Yeah.

You grew up here. You're obviously as Indian as any one of us, but that everybody knows. And, you know, I think that's very inspiring. But talk a little bit about India from the angle of Whisper. Yeah. Right. You're here. I know for a fact when you just mentioned again, one of those three things on a Friday, India launch, you're starting to build a team here. Tell us a little bit about Whisper's plans for India and again, why? Like I asked this question over this last year in between the AI summit, there's so many of these AI founders coming to India. There's like rest of the world to also go to. So remove the Indian out of you and tell us about like Whisper in India.

The question is why not? So this question comes up a lot because I'll tell you the mindshare of both people in India and outside of India. Has been that Indians don't pay for products. That is what I used to also think about a year ago. But the more I dive deeper, the more I realized that was fundamentally wrong and it was from a different era. And so I'll tell you what the key things are when I look at this. So number one, India is actually one of the fastest adopters of AI in the world.

Most products, that's true.

In the world. Which is actually, if you think about it, it's insane.

Yeah. And let's be very specific. I know what you're saying, and just for the audience, it's the fastest growing, which means every time people confuse it with this idea of like, oh, we have 300 million users. No, it's not the absolute. The rate of growth of AI product adoption and the rate of consumption of AI products is one of the fastest in India.

Exactly. And for every single company, like at least by number of users, India is like in the top 3. Always. The second thing is Indians actually pay, but Indians pay when they get a good amount of value, right? So this is a very value-conscious country, and so they're not going to pay for random things that they use as a one-off. And so as a product, if you want to build for India, you need to show that you provide them a tremendous amount of value. And so a very counterintuitive fact for Whisper is The conversion rate in India from free to paid is the same as that in the US from free to paid.

Wow.

And now again, India has been scaling really fast over now and—

And you're already in the millions of users in India?

Millions of users overall.

Overall, not paid.

India is going to get there soon. And so that was the second—

And is it your second largest market?

India is our second largest market overall.

By paying customers also?

By revenue and number of users.

Wow.

This again, counterintuitive.

Who would have thought, right? That for like, like if you tell like the average Indian for a mic, you'll pay. I mean, just think about it.

No, exactly. And so the third thing that we found is Whisper is actually the perfect product for India because India is a voice-first country. Instead of messaging, we love to call. We are the country that presses the voice button on every single app, be it Google Maps or everybody else. In the US, people don't do that at all. And for a country that loves to speak, for a country that speaks in so many different languages, the keyboard's actually the biggest hindrance for us to be able to express themselves, use any other product. And so the thing we've seen about India is the habit formation. Is so much easier because this is how people have wanted to use their devices this entire time. And now finally, that is something that's built for them. And so a big focus inside Whisper also is making this product be the best AI product for India. What are the things that we can do for that? That is why I've come here very often. I spend time with a lot of our users here to deeply understand, okay, what are the things that are missing that we truly need here. And one thing that came up earlier was Hinglish, for example. So Hinglish is how we all talk to friends and family in all of North India. And no AI company till today has solved Hinglish. You solve English, you solve Hindi, there's one language for each. And for me, it was at this point, it was like, if I don't build Hinglish and we launch in India, I'm going to be embarrassed. I can't not have that and come here. And so we actually pushed our ML team to build Hinglish into the model. And for reference, this is a really hard problem.

Extremely.

That no company has solved yet. Not Google, not Apple, not everybody else. Switching between two languages is insanely technical. And so our ML team, they would be like, Tanay, no promises. You're telling us to solve a completely unsolved problem in 8 weeks, but they actually made it. And I love it. And when people here use it, the best thing I feel is they feel like, you know, this is a product for Bharat.

It was absolutely the aha moment for me personally. Yeah, that was it. Like, I, when I first, before, way before we met, when I first got acquainted with the product, that was, that was the aha moment, which is, uh, somebody told me, again, a founder friend told me, hey, listen, you've got to check out Whisperflow. I downloaded it. Initially, of course, as consumers, you behave yourself, right? Not talking in proper English like we are speaking. And then of course, one of the days I remember I was in my car and there was a lot going on that day, right? You lose your distress, you lose discipline. So I just started talking to it and I didn't realize I mixed Hindi words. I didn't realize what I was saying and it got it perfectly right. Yeah, I remember that was the real unlock and the aha moment. I can't even imagine how hard it must have been to get that right. And I'm almost tempted that the next time I'm going to take you up on this, that I'm out in SF or we bring some founders, we'd love to sit with your ML team.

Yeah.

If they have a little bit of time, because it's a problem that just, to your point, everybody has tried it. Global, US labs, India labs, largest companies, thrown a lot of capital, but super hard.

Yeah. Awesome.

What's next? What's the rest of 2026 look like for Whisper? We are already 3 months into your life, right?

I know.

1 quarter gone.

3 months pass fast.

Yeah. So you have 35% of the year left.

Yeah. Lots to do. We're launching a number of new product surfaces, which is primarily all focused around— now Whisper's built trust with people where people just blindly trust it to write on their behalf. And now people wanted to do things for them. This is a request we've been getting from a lot of customers.

Add me to that queue as well.

You're there as well. Because with Siri, you don't want that because you don't trust Siri to do things. Well, Whisper, we've built the trust. And so now we're going to be shipping a lot of actions. And the key thing there that we're building is really looking at, okay, what is something in your life that you do multiple times in the day that just feels like grunt work? Those are the specific things I want to take and solve for you. So it's a very different way to build these kinds of actions and workflows into the product. It's not going to be the basic, like, make a calendar invite. Nobody cares about that. It's solving those real problems. The second thing is now we're starting to bring Whisper to also a lot of companies. Like, we have millions of users, but now we also have 5,000 enterprises that use Whisper. A lot of the Fortune 500, a couple of the FAANGs. Have Whisper deployed to a lot of their employees. And that is going to be, I think, a big thing for us because what we see in these companies is actually insane. You go into a floor of one of these big FAANG companies and you will see people in the company have a mic on their desk. Again, these United Nations mic that we first talked about.

I need to get one of those. That reminds me.

It's phenomenal. And everybody's speaking to their computers, the entire office. It has fundamentally transformed how companies function, how individuals function. And that is the change we're starting to see company after company where the same way, you know, many years ago computers became a thing that every desk had. Now it's this mic that's becoming a thing that every single desk has.

Awesome. On that note, we're going to install some of those in our office. We're going to—

I'll get you some Whisper branded ones.

Would love it. That's awesome. Why don't we get some of those for just like, you know, we can distribute it to founders in India, emerging startups, we can do some giveaways. Maybe we can chat a little bit about that.

More than happy to. We'll talk more and we'll set that up.

Yeah, so we will definitely do that. But Tanay, it's been awesome talking right now and just hanging out this week. Like I always, I say this consistently, people have been asking me all the time about you and just Whisper. And I say this that look, it's rare to find so much consumer love and founder and team depth in a solution or a product or a problem. And I think you're really bringing both of those. So more power to you, more power to you to build a great product. And of course, all of us from a Bharti or Selfish India angle are proud of some of the work that you're doing. And many of us hopefully are already Whisper users, who aren't will check it out. And yeah, waiting till India becomes possibly the largest market for Whisper.

That is where I want to take it.

Awesome.

Tanay Kothari. This is awesome, Akhil.

Thanks.