Wait, can this Yale dropout beat Apple ?

Manav: Today on the show we have Dawson Chen — his company is called Martin. If you've seen Iron Man, you might remember Jarvis from that movie — he's building the real-life version of that. I'm really excited to have you on the show, Dawson, how are you doing today?

Dawson: I'm doing great, yeah, thanks for having me.

Manav: Can you give a little intro about yourself — what were you doing before, how old are you, and what are you working on now?

Dawson: I'm Dawson, I'm 20. Before starting Martin, I was a student at Yale — last summer I dropped out to start Martin with some friends, and for the past year we've just been doing a lot of product work, trying to build basically a better Siri, powered by a large language model with much deeper integrations. It's been a lot of product work, and we recently did some launches we're really happy about, and we want to get it in front of as many people as possible.

Manav: In simple words, Martin is like an AI agent you can talk to — if you tell it to send an email or a text to someone, it actually does it. A lot of times people try that with Siri, and it just doesn't happen.

Dawson: Right — the problem with Siri is it's only very shallowly integrated with a handful of Apple products. What we want to do is integrate deeply with all of your productivity apps. For example, with our phone integration — you can actually call Martin over the phone, and Martin can call your friends and talk to them on your behalf, or even call strangers to ask them questions or book a reservation, that kind of thing.

For every integration we build, we try to make it genuinely deep. For calendar, you can view, add, edit, and delete events with voice commands — text Martin to add or delete an event, and Martin can text your contacts and coordinate scheduling full events with them. So every integration is meant to be deep, and we're not limited to a single platform — we support Apple Calendar and Google Calendar, with Gmail and Outlook coming soon. You can text Martin, speak to him in our interface, call him, WhatsApp him, or email him — we try to be extremely well integrated, and that's really our specialty.

Manav: That's really amazing. How did you come up with the name Martin?

Dawson: Martin was actually the name of my robotics mentor in high school — we really liked his voice, so the Martin voice is loosely based on his.

Manav: Before Martin, what were you doing — is this your first ever software product, or did you build something before?

Dawson: I grew up hacking together websites and apps with friends in middle school and high school — we built things like a bell-schedule app to track your classes. In college, I built a lot of side projects too. One was called Campus Guide, basically Airbnb for college tours — you could book a tour with a current student, they'd meet you on campus and give you a full tour for about $45 an hour. I also launched a social app called Flikk — every morning you'd get matched with a new friend within a mile of your location, and you had to take a photo together within the next 24 hours.

Manav: That's really good — it kind of breaks down social anxiety.

Dawson: Yeah, it was something I genuinely wanted during school — all I really wanted was to hang out with friends and make a bit of money on the side, so both of those products were really built for myself first. I also only ended up going to school for one year — I got pretty bored of my classes by the end and really wanted to do something bigger, and do it full time, so that was the perfect opening to start Martin.

Manav: What was the process like getting into Yale — did your past projects play a role, or was it mostly SATs and traditional stuff?

Dawson: I'm not sure my hacker projects really factored in much. I went to a fairly competitive high school in the Bay Area, so we all worked pretty hard, a lot of academic stuff generally.

Manav: You were at Yale — did you find it boring, or useful? Why did you decide to drop out, when most people would kill to get into a school like that?

Dawson: I genuinely enjoyed my one year at Yale, but by the end I felt this real, almost dire need for more adventure. I think the first semester of college, for anyone, feels like this huge adventure, but somehow it got dull pretty quickly. I could basically picture my next three years — a lot of recruiting, a lot of classes, the same rhythms repeating — and I wanted something new. When I came home last summer after freshman year, I kept telling everyone, "I'm going to drop out, I'm going to drop out," I just needed some kind of excuse. Once we got funded for the startup, it was basically a no-brainer.

Manav: I want to get into your thought process for coming up with the idea for Martin — were you scratching your own itch, had you been thinking about this for a while?

Dawson: I'd been a heavy Siri power user for a long time.

Manav: That's rare, by the way — everyone I talk to barely uses even half of what Siri can technically do.

Dawson: Right, throughout high school I was genuinely frustrated with how bad it was, but I kept using it anyway, because I just love voice as an interface, I find it really compelling. That was the seed of the idea. Large language models were a fairly new thing at the time — this was about a year ago — and it felt like a great opportunity to push the state of the art forward a bit. There were also much better voice models emerging, better text-to-speech, better speech-to-text, and we thought, why not bring all of that together, since nobody had really done a great job combining it, and then integrate as deeply as possible with everything.

We were also frustrated that Siri has such a limited number of integrations — really all it can do is help you move through your iPhone a little faster, but it doesn't touch your email, your Google Calendar, doesn't let you have a full conversation with someone else, doesn't touch Slack, Trello, Notion, any of that. Martin is on both iOS and Android now — well, we're not on Android quite yet, but for iOS users we try to cover Google products, Apple products, and Microsoft products across our integrations.

Right now, the action items are things like making a phone call, playing a podcast, calling or texting a specific person, setting a reminder. The two main categories we started with are scheduling and communications. Scheduling covers reminders and managing your Google and Apple calendars. Communications ties in closely, because the biggest use case there is having Martin text someone on your behalf and coordinate back and forth. You basically give Martin a "mission," and he'll reach out to that contact from his own phone number, converse back and forth to coordinate whatever the task is — and during that conversation, he has access to your calendar too, which ties directly into the scheduling use case. Those are our two main areas right now, and we're just about to launch phone calling — we actually pushed a beta version yesterday, and we're going to roll out full "Martin calls people on your behalf" functionality very soon. That's really the last piece of our communications MVP.

After that, we're going to focus more on email — making Martin able to coordinate and communicate over email, in addition to texting and calling. Soon after that, I think we'll move into document and information management, which would include Slack and Google Docs integrations.

Manav: What's the business model — is it like ChatGPT, where you pay $20 a month and this voice assistant is just active all the time?

Dawson: We're still pretty early in figuring out the business model, but right now we charge $30 a month, with a free trial.

Manav: Do you believe that a decade from now we'll be completely hands-free, just talking to our devices — do you think that actually happens?

Dawson: I don't think visual interfaces are going anywhere. I think audio interfaces have honestly been done pretty poorly over the past decade, though it's definitely gotten a lot better. Like the movie Her — in that movie, everyone's only using audio — and I find that a little hard to fully believe, because some things are just genuinely better typed. If you're coding, I'd much rather type than speak my code out loud. That's part of why we're not purely voice — you can text Martin via SMS, via WhatsApp, or email him, depending on the scenario. Say you get an email listing ten upcoming events, ten Zoom calls, a conference coming up with ten sessions in a day — you can just forward that whole email to Martin and say, "put all of this on my calendar and remind me five minutes before each one," and he'll set up 20 calendar events and 20 reminders from that single email. Trying to do that entirely by voice would take forever.

Manav: That's amazing. What's the ultimate vision for Martin — where do you want to take this?

Dawson: Honestly, the dream is basically the same dream Siri, Bixby, and Google Assistant all originally set out to achieve — a genuinely personal, proactive AI assistant. I think the key to that is deep integration with everything — the assistant needs to be able to use your own productivity apps as competently as an actual human virtual assistant would, and it needs to know you as well as a real assistant would, all your preferences. Personal, proactive, and well-integrated — if you have those two things together, you get a dream personal assistant, essentially Jarvis.

Manav: There's this general myth that voice assistants are secretly listening to you even when you're not actively talking to them — like people saying "Amazon somehow knew I wanted to buy this and started recommending it." Do you think that's a myth? You're an engineer, so you probably understand this better than most.

Dawson: Some of them technically have to — Siri has to be listening for the "Hey Siri" wake phrase, so it's technically listening at all times in some limited sense; say "Hey Siri" with your phone screen off, and it'll wake up. For us, with Martin, we specifically kept it so it doesn't listen when it's off — you have to open the app, or use a shortcut. We have one where you double-tap the back of your phone and Martin activates, or if you have an iPhone 15, you can hold down the Action Button and it'll launch Martin. You can even set up a "Hey Siri, get me Martin" shortcut. So Martin doesn't listen unless you've deliberately activated him through some specific action.

Manav: That's super cool. Have you read or watched anything that felt genuinely life-changing?

Dawson: What definitely inspired me was reading a bunch of Paul Graham essays while I was at Yale — that's actually what made me want to apply to YC. I just started reading through his entire blog, and every single essay was great, especially the older ones from around 2010.

Manav: How did you get into YC?

Dawson: We did YC last summer. The application itself is actually pretty short — I think anyone can apply, it's the same process for everybody, you just apply, and if you get through, there's one interview.

Manav: What advice would you give a 20-year-old founder still in college right now who wants to start a new startup?

Dawson: I think college is primarily a place to find your identity, and figure out who you want to be around. I got far more of that out of college than any specific academic thing. It's pretty clear to me that in college you kind of become the average of your friend group, so don't make up your mind too quickly about who you want to be, or write something off just because it seems too hard.

The thing I hear most often from peers who want to start a startup is that they don't feel ready yet, or think they need another five or ten years before they will be. Everyone has their own internal clock for when they feel ready, and you need to build your own confidence and faith somehow. For me, the best way to build that was launching projects — my campus tours website, my social app — and reading Paul Graham's essays. Between those, I think I built up all the confidence I needed.

Manav: Everyone, go try martin.com and try the product, you'll be amazed.

Dawson: Yeah, I think the best way to actually understand Martin, and what we've built, is just to try it — there's so much you can do with it. If you're into productivity, or if you've always wanted a Jarvis, the way I did growing up, it's a genuinely magical product to try. And let me know any feedback — just have Martin ping me directly.