←
Tanay Kothari2,095 posts
Tanay Kothari avatar✓

Tanay Kothari ✓

@tankots

CEO @WisprFlow | IOI medalist | Forbes 30 under 30 | Stanford CS + AI

◷ Joined April 2017▧ 177 Photos and videos
Tanay Kothari avatar
Tanay Kothari@tankots·Nov 12

story time: >4 months ago, we hired ariya rastrow, who grew the alexa speech team at amazon from 0-400 people, as our CSO and started the wispr advanced interfaces lab. >2 months ago, we launched our first speech model, canto 1. we built it because we know nobody wants to make fixes after dictation. >since then: -desktop users fix 13% fewer dictations. -slow dictations, (the ones that take more than a second to appear) are down 54%. -users are dictating 7% more words each day. >yesterday, @voicearena_ai named wispr flow #1 on its dictation benchmark. we chose to invest in a lab because the way people are interacting with their devices is changing. that requires training larger, more capable models. more soon. wisprflow.ai/research/voice-…

Post attachment
Shobhit Banga avatar
⟳ Tanay Kothari retweeted
Shobhit Banga@shobhitbanga·Nov 11

Drop 4/15: Announcing the Voice Arena Dictation Bench - the first open benchmark of AI dictation apps. Not their APIs: the real apps. We played 761 recordings from 87 speakers into each of 10 apps on a Mac, then 5 LLM judges asked if each draft could be sent as-is. Wispr Flow: 79%. Apple: 35% 🧵

2:15
Yu Chen avatar
⟳ Tanay Kothari retweeted
Yu Chen@Yuchen_SigNotes·Nov 8

14 days on @WisprFlow. Switched to Pro today. Here's what changed in my Claude Code logs: Prompts to my agent: +51% Long dictated briefs: 4x longer Time spent with my agent: +80% The real unlock: I mapped record and send to the two side buttons on my mouse. One thumb tap to talk, one to send. One of the best product experiences I've had in a long while. Smooth like butter. Keep it up! @tankots

Post attachmentPost attachment
Zach Moskow avatar
⟳ Tanay Kothari retweeted
Zach Moskow@zachmoskow·Nov 5

laggy, buggy, slow. misses first words. not bad at all just not there yet

Anvisha avatar
⟳ Tanay Kothari retweeted
Anvisha@anvisha·Nov 4

We just made a 5x cheaper Higgsfield. Voyager gives you AI video models like Seedance 2.5 at half Higgsfield’s price. More importantly, it cuts wasted generations. Users get final outputs in 2–3x fewer API calls. Half the price per generation. A fraction of the generations.

0:49
Tanay Kothari avatar
Tanay Kothari@tankots·Nov 3

my dad recently asked me what he should actually use an ai assistant for. i’ve spent years building in the ai space, and i honestly couldn’t answer him. he recently started trying out ai assistants like instinct, muse, dots, etc. for most of them, you can send a text, and they'll go off and do different tasks for you. the catch is you have to tell them exactly what you want. my dad doesn’t really have a list of jobs waiting to be handed to an assistant. he also doesn't know what they're capable of, so he doesn't know the right place to start. people in tech don't have this problem, because we use these tools every day. we connect them to fifty apps and hand them half our tasks. but people like us are maybe 1% of the world. most people just want something that can solve one clear problem for them, not fifty. i’ve seen this problem before. when we started wispr, some of the early dictation tools in the market worked decently well, but they were complicated to setup and use. anyone like my dad was never going to do that. flow took all the setup away and made it something you never have to think about. that’s what made it usable for people like my dad. before you build technical products, you should be confident that a parent, grandparent, or someone non-technical can understand the problem it solves. that's how you build tools for the other 99% of people. - written with @WisprFlow

Tanay Kothari avatar
Tanay Kothari@tankots·Oct 28

these are the 4 most important questions to ask when deciding on a co-founder: (these have worked wonders for @SahajGarg6 and i) >do you care about working on the same problems? if not, you pull the company in different directions. disaster waiting to happen. >do you share the same values? ours were integrity and being customer first. that meant no ads inside our products, and making product experience the #1 thing. >do you want to work with the same kind of people? we chose a small, extremely capable team of people who inspire us professionally, and are a joy to be around. >do you complement each other? if you make decisions fast, you might want a co-founder who decides more slowly. if you're shy and don't like talking to a lot of people, that's fine, but you probably need a co-founder who's more extroverted. these will help you pick a person that you can build a massive company with, and enjoy it along the way. — written with @WisprFlow

Tanay Kothari avatar
Tanay Kothari@tankots·Oct 28

i'm often asked, "do you think i have pmf?" my answer is always the same: "if you're asking that question, you probably don't." when you hit pmf, you won't have to ask. people will love your product so much that you'll hear it directly from them. we found this out firsthand at wispr when we heard that people were buying macbooks just to be able to use flow. you can't fake pmf or people loving your product, and it will be obvious when that happens. until then, keep building. ship, collect feedback, ship again — written with @WisprFlow

Gaurav Vohra avatar
⟳ Tanay Kothari retweeted
Gaurav Vohra@gauravvohra·Oct 27

Tanay Kothari (@tankots), CEO of @WisprFlow, has the cleanest definition of brand I've heard: You shouldn't just think "voice" when you hear "Wispr Flow" You should think "Wispr Flow" when you hear the word "voice" Sports shoes → Nike Coffee → Starbucks Voice → Wispr Flow

0:33
Tanay Kothari avatar
Tanay Kothari@tankots·Oct 28

when we raised our seed round years ago, i actually told investors exactly how we *could* fail. i'm generally a no bs person, so i wanted to lay all the cards out. the pitch had 3 parts: -here's what we think this could become -here's what we know so far -here are all the risks for how this company could fail i told them in a meeting, "you could tell me 5 risks you see. i'll tell you 100, because we've done the work to understand all 100." we raised $4.5m a couple of weeks later, and the journey started. investors aren't stupid. they hear pitches all day, and they can tell when a founder is hiding the downside. so if you're looking to raise: don't oversell the upside, and definitely don't hide the downside. honesty wins. — written with @WisprFlow

Tanay Kothari avatar
Tanay Kothari@tankots·Oct 27

2 years ago, i had done zero marketing. i'd been an engineer my entire life. when i told a founder friend we were building a consumer product, he told me, "tanay, you don't have that consumer dna. you don't know how to build a brand. you shouldn't do this." it sounded like a fun challenge. game on. 2 months after we pivoted Wispr from hardware to Flow, our first launch did millions of views and hit #1 on product hunt for the day and the week. what i've learned about marketing since: 1/ take almost no inspiration from others. standing out means doing things your own way. be unique and bold in your decisions. 2/ pick your first users by who will cause PLG. hundreds of avatars of people loved wispr flow. we focused on 2. developers, because they post about tools they love everywhere. founders, because they roll them out to their whole team. 3/ make every user feel it was built for people like them. most tools assume american english. in wispr you can pick british english, swiss german, or simplified vs traditional chinese 4/ build small moments that make people feel seen. in india, after you've spoken 1,000 words in an indian language, a banner says congratulations in that language, with the flag 5/ do pr even if you can't measure it. if i put its roi in a spreadsheet, the column would be empty. but asking what i'd pitch a journalist about every new feature gives us a lot of our ad ideas 6/ make the message something you can see, touch, feel. have fun with it. we said we'd kill the keyboard, so i sawed one in half in our office for a launch video 7/ treat word of mouth as a science, not an art. after we hit product-market fit, the first thing we optimized for was getting every user to tell their friends turns out you don't need consumer dna or to be the most creative. you just need to treat marketing like a science.

Tanay Kothari avatar
Tanay Kothari@tankots·Oct 26

most ai products right now are built for the 5% of people who are tech savvy. my dad is never going to write a system prompt. my grandfather doesn't even know what a prompt is. that means it's the best time ever to use ai and build products for the other 95%.

Tanay Kothari avatar
Tanay Kothari@tankots·Oct 26

the 4 things i look for in every hire: 1/ a strong first-principles thinker. someone who works out why something is true instead of repeating what's usually done. twenty years of experience often teaches you what's "supposed" to work. i want people who check whether it does 2/ high cpu. some people call it high iq. i think of it as how quickly you can absorb new information and have a conversation about it. at a startup, what you need to know changes every few months, so speed of learning matters more than what you already know 3/ high eq. it shows up as self-awareness and low ego. the biggest difference i see between an extremely smart intern and an average one is how quickly they learn from feedback, and you can't learn from feedback if your ego gets in the way 4/ obsession. is there something you're deeply, madly obsessed with? i've been obsessed with the same question since i was 10. what does the next generation of computing look like? the people who do their best work here usually have a question like that experience isn't on the list. we get applications from staff engineers and principal scientists all the time. one of our best engineers was a 20-year-old with no fancy resume who dm'd me on a saturday, came into the office 2 hours later, and shipped a full feature by sunday morning. aptitude > experience — written with @WisprFlow

Tanay Kothari avatar
Tanay Kothari@tankots·Oct 23

every leader's job is to make sure the hardest thing their team deals with is the work itself, not bureaucracy or feeling unappreciated. when i was right out of college, i learned this the hard way. i'd never managed anyone before and i was running a 25-person team, most of them older than me. in my first month as a manager, i messed up so badly that someone on my team cried and almost quit. this was a huge turning point for me. i knew in that moment that i needed to learn how to be responsible for people. that means giving them clear guidance and mentorship. it also means staying out of their way a lot of the time. here's 5 practical ways i do that now: 1/ one day a month, i shadow someone on my team for a few hours. then we map what's blocking them and fix it together, and apply it to the rest of their department 2/ in a new hire's first 2 weeks, i walk them through how i prioritize and why. most companies fail to scale because new hires don't understand how the founder thinks, or worse, the direction of the company 3/ with senior people, i define the problem and stay out of how they solve it. early on at wispr when we needed better ads, our cmo built 100 of them himself in figma, tested them, then brought in a team to scale it 4/ when someone needs to grow fast, i give them one thing to own and 30 minutes with me every day. this allowed a 24-year-old on our team to go from small experiments to managing engineers in 8 weeks 5/ i hand off what i used to control. i was deeply involved in every marketing launch we did for a long time. for notetaker, i told our cmo "i trust you with this. we've worked together long enough." taking care of your team is priority number 1 — written with @WisprFlow