The Truth About Voice AI: Wispr Flow’s Tanay Kothari on Habit Formation and Building His Own 'Jarvis
21 May 2026 · 25 min
Tanay Kothari built Whisperflow after seven failed AI assistants. His voice-to-text tool interprets intent and tone across 100+ languages. Users hit send within half a second of seeing output, showing they trust it enough to use it dozens of times daily.
Chapters
But it does make you who you are, and part of the reason why I push myself so hard, part of the reason why I work 20 hours a day most days, part of the reason why there's a bed in my office is because of this imposter syndrome that's pushing me.
Welcome to Tech Talks, hosted by myself, David Savage, and powered by Nash Squared.
On today's show, we are peeling back the layers of what it truly means to build a company that not only serves a problem but fundamentally changes how we interact with technology every single day. My guest is Tanay Kothari, the CEO and co-founder of Whisperflow, one of the fastest growing AI companies in the world right now. When you think about voice dictation, you probably think about frustrating tools that come built into your phone or laptop, the ones that make endless mistakes and sound fairly robotic. But Tanay and his team are doing something fundamentally different. They've built a voice AI that doesn't just hear you, it understands your intent, your tone, and even the structures of your rambling thoughts and puts them into perfectly crafted messages, whether it's for a formal email or or a casual text. But the journey wasn't straightforward. It's been an odyssey through 7 previous product iterations that didn't work, and it was a quest born from a childhood dream after watching Iron Man to build a real-life Jarvis. It's driven by powerful, perhaps even counterintuitive, insight that the sexy innovations that often are the things that people want, well, maybe aren't the things that build the best businesses. So we're diving into Whisperflow's story and how Tanay thinks about business. I think there's plenty to learn.
So today I'm joined by Tanay Kothari, the CEO of Whisperflow. Tanay, thanks very much for giving up some time this morning. It's 10 o'clock in the morning in San Francisco, right?
It is. Great to be here, David.
I'm pretty sure it's 8 hours behind, although every now and then around the clock's changing, it sometimes changes.
Yeah. Yeah, no, the UK and US daylight savings don't match.
No, they don't.
It does a thing in between.
Yeah, it catches you out. It's sneaky like that. Anyway, enough about— don't talk to me about weather and daylight and whatever else. It's a Brit's favorite topic and it's not one we're here to talk about today. Let's talk about Whisper and first of all, ask what you are building.
Of course. So I'm the CEO and co-founder of Whisperflow. And Whisper is a voice dictation tool that unlike every other that's existed for the last 20 years, actually doesn't make mistakes. What we cared about was the fact that you speak differently than you write. And so Whisper hears you and it writes what you would have written, including being formal in emails, casual in texts, getting people's names right. Structuring your rambles, letting you do that in over 100 languages. And it's one of the small reasons what has made it one of the fastest-growing AI companies in the world today.
And what are the typical use cases? I'm assuming it's something that people are using in work settings, but also, I don't know, perhaps enthusiasts, hobbyists. Where are you pitching this?
You know, people surprise me. It is used a lot by See, the average manager who spends their life in email and Slack responding to messages all day long, the shocking thing is the average person spends 5 to 6 hours a day typing on their phones or computers. And with a tool like Whisper, that actually drops in half. And so they save 2 to 3 hours a day, which if you add up in a year is huge. And so this includes everything from text on iMessage or WhatsApp to your Slack or email. People are using this with the latest wipe coding tools like Cursor and Lovable to build products. And with those experiences, what we've learned about AI is the longer prompts you give, the better the model is going to do, the better responses you're going to get. And people are extremely lazy when they type. And so if you give them the ability to just yap to something and they just go on and on and on, the models actually do a fantastic job and they're really happy with the outputs of the LLMs, be like ChatGPT or whatever else they're talking to.
I would imagine that it's more suited to someone replying to an email where it's the kind of typing that's not fun. And let me explain that. I am someone who likes to write. I'm someone who quite likes writing copy, so therefore I can obsess over whether or not I've used the right words to convey the meaning, and I'll write it 4 or 5 times. And I actually quite enjoy that process, but I hate replying to emails. Yeah, in my mind. But I suppose everyone's different, right? But that's how I could imagine something like this really coming into its own.
That is generally how I think about what we're doing as a company. Like, as a company, our main focus is we want to take out the grunt work in your life. All the things that you would rather not spend time doing so that you have a lot more time to do the things that actually give you joy.
Where did the idea come from, if you don't mind me asking?
So way, way, way back, back in 2008 when the first Iron Man movie came out, I wanted to build Jarvis. And so me and a friend of mine built back then what was one of the world's first voice assistants, right, before Siri, Alexa. Grew to millions of users. Google shut us down. And then fast forward many years, this idea came back again where when I look at people on their computers or phones, it feels so mechanical and so effortful. And to me, technology should be something that should feel as effortless as talking to a close friend. And so that was the birth of Whisper. And it was actually one of the 8 products that we built at Whisper before before launching the current one that is out there. And this one primarily came when we just sat beside people and saw what they did day in, day out. And we realized that people do not need an AI assistant to book an Uber. They don't need an AI assistant to book a flight. You don't do that that often, and it's, it's fine if you have to do that manually, but you do type a lot. And if that is something that we can really solve, we give people back a lot in their life that they wouldn't have had otherwise.
So it's the 8th product. So was it 7 iterations of a similar idea, or were they— were they very—
7 iterations were completely different. The 7 iterations had to do with building AI agents that did things for you, uh, and it felt really cool, it felt really sexy, right? It would, it would book reservations and send a message to your friend knowing that the reservation was made. You could be like, hey, David's coming to visit, find a cafe, make a reservation, let him know about it, book my calendar, and it would do all of that. But turns out it's not all the sexy things that should make it out as, as products, because that's oftentimes not what people want. Yeah.
Okay. So what was wrong with existing voice dictation tools? Because there are plenty, of course.
Yeah.
Freely available on your, on your phone. So what didn't quite work about what's out there already?
Yeah, they come with your phone, they come with your laptop, they come with most of the, you know, your Google Doc and Microsoft Outlook out there. But if you have ever used them, there's some things you might have noticed. The first one is it makes stupid mistakes where every time you try to use it for something, there's at least one mistake that it makes that you need to go and fix. So you can never blindly trust it. The second thing is it sounds a little robotic. It tries to make your grammar perfect, and while doing that, it loses the authenticity of your tone. And finally, the biggest issue is you speak differently than you write. You speak in rambles. Your writing is a lot more structured, and you just have to go and do that manually anyway, because all dictation systems until today write everything you say word for word. Turns out that's not what people want. What people want is, I want to speak and I want something to show up that is exactly what I would have sent in my tone for this given context. And once you realize that everybody else had been solving the wrong problem until now, you start to look at things very differently, which is what we did. And for the first time, Whisper, something that actually solves that. And the one metric that we measure is actually a really interesting one because, uh, uh, this is very unique to Whisper is how much time does it take for a person when they see Whisper's output to press send? And it's half a second. It's at a point where people aren't even reading what Whisper writes anymore because they're just like, it's probably right because it has done that hundreds of times before.
I love the ambition.
Yeah.
I would say in a corporate environment, that feels quite risky, right? Say you're in an enterprise and I trust it's probably right.
Yeah.
I don't know. I'd imagine the legal department isn't thinking about it that way.
That is what I thought. That I am just as surprised at seeing this number as you are. This is not ambitious number. This is the reality. And I thought we'd get here in a year or two's time, right? Because people are particular about how they write. Turns out it at least does a better job than what they were doing before.
I get it. I think it's a great idea.
Yeah.
I love the fact that this is the 8th and you had 7 different iterations of different businesses and you had that kind of idea of watching Iron Man. Like, so much there is really to be admired. So let's pick into that because I think anyone listening is always— we always get presented with a success story at the end, right? We very rarely go through the journey of what quite didn't work. And I certainly don't want to fetish failure. That's by the sounds of it, this is iterations and pivoting and learning. So let's frame it like that and ask you what the most interesting learning moments from that journey, from that first, I want to build an Iron Man Jarvis through to what you have today are. What were the real kind of moments that helped clarify how you build a tech business as you went.
So the way, the way we built things at Whisper was we had an idea for a very clear problem that we wanted to solve. Then we quickly make a prototype for it and we start using it internally, right? And the one thing that was very clear from us from day one is if this is something that we don't use, nobody else is going to use it. So we're good, essentially, uh, you know, guinea pigs. To test our own software on. And the first 5 ideas, if I tell you, like, they sound like the coolest things ever, right? People call them— this is back in 2020, 2022. This is before AI agents was a word. This is before, you know, this is when ChatGPT had just launched. We're talking way back when in the ancient days. And the thing we realized was all the things that we built that sounded sexy, I would use it once, twice, and I personally would forget that this product existed. And that was because what we realized that when you're building a new consumer experience, it is all about building habit. It is all about building habit. And so when you think about what leads to habit formation, it is doing something multiple times at, in a particular situation. And so how many times do you book an Uber? It's like, not that often. How many times are you setting calendar invites? Not that often. How many times are you booking flights? Like almost never. And so that got us to the thought of what is something that we can build that people use not once, not twice, but 100 times a day? 'Cause that is gonna build habit. And that was the source of this 8th product, which is voice dictation. It doesn't sound sexy, but that's fine because once we built it, I know I was using it for every single message I was sending. And I had just built a habit of it where I used it nonstop for months. And whoever we gave it to also used it nonstop for months. And that to us was the sign where this learning got cemented very deeply in my head, that habit formation comes from frequency of usage. You can build all the cool things in the world, but if you can't build a habit, your company is just gonna die. Yeah.
There's something I love about this, right? Which is, it's a really good message. 'Cause I think when you do start this idea of being an entrepreneur, a founder, everyone has very high-minded ambition to change the world.
Yeah.
And I was certainly— this sounds awful because it sounds by kind of insinuation that I'm saying that you're not changing the world. No.
But there is a realization that, no, you need to build something that people want.
And that perhaps is lost on a lot of people initially. That is not simplicity is the driving thing.
Yeah.
Yeah. So look, how do you get it out into the hands of people? Because as we've said, these tools exist on people's phones already. I do a lot of typing. I spend a lot of time on my laptop and on my phone. I work in tech. I hadn't heard of Whisper to this point until interviewing you. So how do you get word out to people who might be users?
So the thing we realized is, David, if I come to you and I tell you, David, I have an incredible voice dictation product, I think you should use it. In your head, you're going to be like, I tried Siri, Siri sucks. I'm not going to use this thing. And it's okay. I wouldn't be offended if that was the thought you had.
I am an Android user. I generally think that, I generally think that iPhones aren't great. I have a MacBook. I love them, but iPhones, no.
Anyway, no, that's a different conversation. Only point for you. Uh, so, but what we realized is if somebody else tells you, David, I started using this tool and it changed my life. And this is somebody you respect, right? You would probably be curious and you would check it out. And maybe that's why you are here on this call with me trying to figure out why you heard a lot about Whisper from possibly multiple people. And that is what we engineered. So to me, the best products spread because of the users using it. That is the same with all the best things in the world, right? You get a recommendation from somebody to go eat at a really delicious restaurant or to buy the softest hoodie or whatever else. And the way you generate that referral is you give the user an aha moment. And that is what we engineered into the product. So the product was just as much about you know, everything else that you think about while building a software product, as it was about how do we get people to the aha moments and how many aha moments can we generate. And so when you use the product, you're a new user, you start using it, within 30 seconds you get to the first aha moment, which is people's first dictation, where they say something, they ramble away, and within a few hundred milliseconds you see whatever you said perfectly on your screen, formatted just as you would have written it. And that blows people's minds. And so what we narrowed down was the time for people to send it to others, which at this point is someone downloads the app, say at 9 o'clock in the morning, by 1 PM they've already dropped it in 3 group chats because of that aha moment. And there's a few more aha moments like these that we have crafted along the user journey. But it's all about this. At the end of the day, it's all about the user emotion that you are really building towards. And so inside the company, when we're even talking about any new feature that we're building, the, the conversation is, here's the spec, here's what we'll do, what is the user going to feel when they see this? What is the user going to feel when they use this? And how do we optimize it for that? And very small design decisions because of that discussion that happens.
Just out of interest, as a kind of a young challenger startup, do you worry about staying cool? And I ask that because you see a lot of memes online of like people taking the piss out of operating systems for, you know, how likely are you to recommend this to family and friends? Like, funnily enough, I don't go around recommending operating systems to my family and friends, but that is kind of the dynamic that you're looking for. So how do you get that secret sauce where people— I know you said about the aha moments and 9 o'clock through to 1 o'clock, But engineering that is still quite a tricky challenge from a marketing perspective and getting that line right.
Yeah, it's not easy. It takes a lot of iterations, a lot of hard work, and I don't worry about staying cool or having that as a part of our company philosophy. Yeah, I If we stop doing that, that would be a bigger symptom to me. And that symptom would be that our team has stopped caring about how our product makes people feel.
Yeah.
That to me is an organizational level problem. It's gonna show up in so many different ways. This is just gonna be, be one of them. And so I would just think about that more from first principles.
You're a reasonably young founder. Is that fair?
Yeah, I'm flattered.
Do you ever get imposter syndrome or not? It's something that we often talk about in the industry, especially when we talk about women in the industry.
Mm-hmm.
Tends to be something that we don't talk about when we see successful young male CEOs, entrepreneurs. You know, you're the kind of person that probably gets lauded, put on stages at Web Summit or CES or whatever else. But from your own personal perspective, do you, you know, you mentioned you're one of the fastest growing AI companies in the world, you're Forbes 30 Under 30. Do you ever kind of grapple with that yourself?
All the time.
Yeah.
Yeah. It is one of the things that is just a constant voice in the back of my head that never goes away. And Part of me really feels like I don't want it to go away because, you know, if there's a— there's a lot of work by Carl Jung on this idea of your shadow self is the part of you that has negative emotions, may bring on negative behaviors, negative tendencies, and you could try to get rid of all of it But it does make you who you are. And part of the reason why I push myself so hard, part of the reason why I work 20 hours a day most days, part of the reason why there's a, there's a bed in my office is, is because of this imposter syndrome that's, that's pushing me. And I'm honestly afraid if that were to go away one day, I might chill out. And I don't want to be that.
So when you find yourself in a room with your peers, yeah, and you're talking to other founders, CEOs of businesses, what questions do you have for them?
I operate in two modes when I meet somebody new. Mode number one is, you know, meeting another founder, in which case all I care about is building an authentic connection. Right, building— being a founder is somewhat of a lonely job. Like, when you are in your company, like, you have to always put on a face of strength. You have to demonstrate stability. You've got to show that you've got your shit together. And it's only with other founders who have— who are in that same journey as you do— they really get you. So that to me is honestly a fantastic time of just a release when you can just be truly yourself and talk about all the things that you're worried about, the imposter syndrome, from everything else. Um, and just honestly open up. And the, the second case is where I'm talking to somebody and there's, there's something that I want to learn from them. And, um, to me, when I think about journey as a, as a founder, right, you are building a company, your company is scaling up in revenue and number of people and all of that. And the thing most people don't realize is you personally need to level up at the same pace as well. And so I always look towards people who are a step or two ahead of me to learn from them specific things about things where I know are areas where I need to improve on. It's like, hey, we're scaling from a 10 to a 30-person company and it feels like chaos. How did you help solve it? Or, hey, we're about to increase our marketing spend from $50K and we're going to spend a few million dollars on things. I'm terrified. I've never spent that much money before. How do you think about it? And I always come from a place of just like, hey, you know, open kimono. Like, you've probably felt this before. I'm not gonna hide anything because that's how you can best help me. And that helps a lot for me to keep up-leveling myself. And it's something that I tell everybody in our exec team to really focus on because If you can keep growing with your company, you're gonna stay on, because if you don't— and that's why you see most C-suite folks have about a 2 to 3 year tenure at companies, because then they're no longer the right fit for where the company has grown to and they haven't been able to grow as well with it.
Do you ever worry that you might not be able to level up at the pace of your own business and then what that might mean for you?
That is the one thing that I do have a lot of trust in myself. That is one thing that I do well, is learn new things and grow. And that is one of the traits that I am most proud of. Everything else might go haywire, but that is one thing that I know that I can keep delivering on.
Look, I think you've been really open, and, uh, thank you for that. Your time has been, um, very valuable to the, to the listeners in the podcast, I'm sure. Let me ask you one last question. You're obviously building an AI tool that you would love people to go and— you said it's an app. I assume it's on Android and iOS.
It is on Mac, it's on Windows, it's on iPhone, and Android is going to come in a few months.
Good.
Yeah.
So obviously people can go check out your product. Give a shout out to another product. What AI products are you using that's making a tangible difference to you every day?
One of my favorite products right now is Superhuman. Oh yeah, yeah, yeah, yeah.
Rahul's a good guy. We've had him on the show.
Oh, fantastic. Yeah, he's been a He's been one of my idols and also somebody I really look up to, been huge fan of the product. And the thing is, right, they built an incredible productivity tool, but most recently I started using some of their AI features as well. And for the first time, uh, an AI tool has written my messages and responded to people just as I would have wanted them to. Apart from, you know, Whisperflow, of course. And so it's just been so delightful, and that aha moment has made me recommend it to at least 50 people in the last couple of weeks.
And therein is the proof that you know that it can work for your business too. And obviously it does.
Oh yeah. Oh yeah.
But it's been a pleasure to speak to you, Tanay. Thank you very much for your time.
Have a lovely day.
Thank you, David. This was fantastic.
Tech Talks is hosted and edited by David Savage. It is produced by Nash Squared, and we have special thanks to Lemsy for supplying music to this show.
More from Tanay
He went viral at 11 y/o—built the world's best dictation app, then raised a $30M Series A. | Tanay Kothari, Founder of Wispr Flow
#115 – Tanay Kothari, Co-Founder & CEO of Wispr Flow: How a 27-Year-Old Founder Is Redefining Leadership, Product, and the AI Revolution
How Wispr Flow manufactured viral moments by personally onboarding 500 users on Google Meet | Tanay Kothari ($56M Raised)