The $2B AI Startup Changing Content Forever | Victor Riparbelli

24:45It's kind of like a microcosm of a business or going to war, like whatever problem, like strategy thing you're dealing with, right? And what games allow you to do is to essentially just simulate things over and over and over again, which I think teaches you decision-making, long-term, short-term thinking. In a way, it's just very difficult in the real world, right? If you're playing soccer on the football pitch, you can do that, but you can't do it alone. And every simulation of like playing a game takes a really long time. You don't always have the ball yourself. Like in a computer game, right?

25:19You can actually sit down and practice while having fun a million times how to allocate resources, how to think about like risk reward, how to like look at you know what's the opponent doing and i think all this is like running a business is kind of like playing a game life you could almost argue it's kind of like playing a game right and i think it's just i think those years for me spending so much time from the outside it looks like you're wasting time right like why the hell are you sitting killing like some imaginary dragon but i just think a lot of the skills that you pick up are very useful because it's all about logic.

25:53It's about being analytical. It's like the world we live in for better or for worse, right? It's like, it's so abstract. Finance is incredibly abstract. Business is incredibly abstract, right? I don't think there's actually that much difference. There's a lot of differences, but I think it's undervalued how much you can learn about being successful in a professional corporate world today from playing these kinds of games that from the outside, it looks like you're just wasting your time. I 100 % agree. And that's like a big thing I have, which is like, you know, I interviewed here the Vercel founder and we talked about this because when we were growing up, computers were so instrumental to kind of like our growing up.

26:35And it's like you said, like now there's all these, you know, anti-computer, anti-iPad things like, you know, how do you draw the line? Right. Because it's kind of hypocritical, you know, of how much computers are important. 100%. I think the main difference, I think that the software makers were less sophisticated back then to optimize for addiction versus for, you know, kind of game craft, you know? Because like, I think they're like, kind of like we're talking to like hardcore games. Those were like really kind of like character building. but like Candy Crush you know on the other hand is not really kind of doing much or TikTok scrolling right and stuff like that so it's I think what they're doing in the computer probably matters as much but the question I have now is like I wonder what is going to be like growing up with ChagyBT you know with like all these skills and knowledge that you get to access it's deeply fascinating and I think every technology trend right comes with kind of like a counter trend as well like there's a lot of things that's very good there's some things that are very bad and I think over time we always like Like I generally am pretty optimistic.

27:41We iron those things out and it ends up having clearly been that positive for society. I think with all the ChatGPT stuff, right? It's so interesting because we're all, I think it's also a technology that's just like so rapidly being adopted by everyone. Yeah. Whereas like social media, for example, like it took quite a while to fall, like your moms and dads started to be on Facebook, right? This is really feels like in like this group of like two or three years, most people in a professional job are now using ChatGPT. some of the things i i think all the upside has been covered very much by other people i think clearly like having a one-on-one tutor to everything you do is amazing um just for my own like when i learn new things i so much easier so much easier right it's it's fantastic the only thing i worry about is i think that's very specific to the models right but these models are kind of offense, right?

28:32Like they kind of tell you what you want to hear most of the time. And generally, I think that's bad. Like when you hire executives, you really don't hire executives that tell you what they think you want to hear. At least I would say most great founders don't. You want a high executive that's going to tell you your blind spots, right? Challenge you on like every decision you make. That's a good executive. A good executive is not someone who just like says whatever you think they want to hear. And when you see a lot of people using GPC for things like therapy, for example, I also worry that a real therapist will challenge you, right?

29:03If you're telling your therapist about some situation where you may be actually in the wrong, the therapist will generally tell you that. With ChatGPT, it's not always obvious that ChatGP is going to tell you that you're in the wrong, right? Maybe they'll just give you an argument for why you're actually in the right and then people will believe it. So that's, I think, an interesting thing of like, how do we change the dial of like, you know how much generally people want people to agree with them it makes you feel good right but but i think what people actually need is like gpt's that agree less with them and the other thing i find i guess maybe more philosophically interesting is if you think of all the relationships you have in your life with like your your partner your kids your friend your colleagues whatever this is probably going to be like one of the most important relationships you have that's pretty freaky to think about right just like how much that gpt know about you and how much will your gpt eventually impact your decision making and your development and so on and so forth and it's one of those things i think happens like slowly but surely to all of us i mean just i think the experience that most people have is the more use these things the better they become and by now i've i've told gpt a lot right it probably knows me better almost than anyone and that's an interesting idea right that that actually like the computer is going to be like one of the most important relationships in your life.

30:22Yeah, that's very different than the past. Absolutely. Absolutely. And so before we get into all the AI stuff, let me go back a little bit to kind of your growing up. So then, you know, you decided to apply for college. You know, were you a good student at this point? Like, were you kind of, you know, kind of into college? Like, how was that experience for you? I was like mediocre. I was okay. I was like probably, I don't know, like 18 percent out of that. So still okay. In Denmark, though, it's a very different, it goes big back to the egalitarian thing, right? In Denmark, it's not nearly as competitive as a process.

30:57It's not like you have to study real hard for your GCEs to get like into university or whatever. But I did go to university. In Denmark? In Denmark. First, I went to like business school and I absolutely hated it. So dropped out after six months. I just thought it was really boring. at that point in time I had no idea I would end up doing business that was not like in my sphere of interest at all um and then I uh was kind of like you know just a bit soulless for a while just work some like you know like normal jobs and my dad was like why don't you do something with computers you love computers like you're so good at computers and it was this really weird moment where I was like that's actually like not a bad idea why have not thought about this before and so my dad took me around to like some different go to like universities whatever and like spend that hour listening what is it like to study computer science what is it like to do all these other things and I ended up studying at the IT University of Copenhagen where I did kind of like a split degree and just kind of like half CS and half business and once I started there I got interested in entrepreneurship because I think I felt that I'm a I have a lot of, I'm very kind of rebellious person, I think.

32:15My instinct is always to disagree with what most people, like my instinct is always like, if someone tells me something, my instinct is always to think about like, why could that person be wrong? As opposed to like, oh, that person is probably right. Very different than the Jentelagin culture. Very different than the Jentelagin culture. You know, I tend to believe that most of the things we believe today are probably wrong. Like things change all the time. And I think the idea of a corporate job to me back then was just like, as someone who wanted to be like a techno DJ, was just like absolutely soul destroying, right?

32:49Like you go into this big machinery, you put this like corporate mask on and you just like go and turn the wheels and like some big faceless corporation. That didn't sound very interesting to me. But I think what I liked about the whole startup scene was that that is very much like you go against the grain, right? You have some belief that few other people think you're right on. You take a chance. and I think so I think that rebellious bit like you know I used to skateboard and like listen to punk music I think a lot of like that energy I think you find in the startup world so that got me very interested and I got very obsessed with that very quickly I did the three years there I did one semester exchange semester at Stanford as well and I absolutely hated it I thought it was a complete waste of time the university or Stanford?

33:30the university Stanford was amazing Stanford was like amazing probably the best like some of the best four months in my entire life but the three and a half years I did it at this Danish University. First of all, let's maybe just, the difference between a Danish university and Stanford University, which I had no clue about until I actually got to Stanford. Danish University, for two and a half years, I worked full time on other jobs, doing startups and working in startups, et cetera. Two weeks-ish before the exam, I would sit down, I'd read a bunch of stuff, I'd write a report, I'd turn it in, I'd still get like decent grades.

34:02Didn't get like amazing grades, but I got decent grades, right? You don't have to be there. There's very little things in between a semester that you actually have to deliver. And everyone has a job on the side. And it really feels like the bar is like, is very low in my opinion. I think the general level of ambition and... I was very, very uninspired to be there, but I did it because, you know, kind of, why not? When I went to Stanford, it was like completely different, right? And I have a sense a lot of American universities are like this, but that was like a full-time, full-time job. Yeah. That's like, you go to class and you have to be in class.

34:40And then you just study for eight hours a day. I was completely shocked. I had no clue it was going to be like this. Did you take computer science classes when you were there? I did take some computer science classes. But I just remember, I mean, that was literally a full-time job. You would study from morning until late evening. And maybe you'd grab a beer in the weekend or something. But that was a full-on experience of really learning something. And I really liked that, actually. and also had this thing in the classes do you call it like when they curve the classes where it's not you don't look at like each assignment just individually you kind of stack ranking for the entire class yeah it's like great curving yeah exactly also very very non-danish right yeah it's basically like it's a competition like if you know if everyone writes amazing papers someone is still going to get like a less good grade so then you get this like competition you have to like find groups of people that you would work with and you kind of trust it because you didn't want to like cooperate too much with everyone else and share your homework because that could actually be detrimental to yourself.

35:37And so what I've met was obviously just a bunch of extremely driven, very ambitious people. I think getting to be in the Bay Area, experiencing like the American optimism when you come with a crazy idea, right? Instead of being like, oh, come on, Victor, like just, you know, why are you even thinking about that? Like, oh, that's interesting, right? Have you heard about this other idea and this other thinker over there? And I think getting that feeling was just, was amazing and changed my life for sure. And, you know, it's interesting, you know, I went to Stanford for a few months, too. And I remember when I went to grab a beer with like one of the professors.

36:14And I was like, oh, tell me, like, how does it work to give a class here? And he was like, well, you know that every question and every problem set and every test, we measure the performance across cohorts. And all of them map out to different concepts that we're teaching. And we try to then change the classes and the problem sets to improve the students' retention in that concept. So every class, we're getting better and better and better at teaching those concepts to students. And I'm like, wow, this is so different than Brazil. And I think a lot of universities in the world, the amount of care they have that the student actually learns versus just passing through.

36:54You really felt like, I really felt like this was a place where people really wanted to be there. they worked super hard they also obviously paid lots of money to get in and I think that energy and also like all the professors the teachers were just incredibly amazing people right like remember like good teachers great teachers I think you had a class and like you know very like one-on-one machine learning stuff there's like one of the google guys like ran the ml division back then I remember like absolutely flabbergasted, especially when you come from Denmark, where, I mean, there's lots of great professors in Denmark, but it is still just different.

37:31You feel like you're being, I had an entrepreneurship class, right? And you're taught by someone that has actually built like a product that you know yourself, you've used it yourself, you know, the story of the company. And so I think that was such a formative thing for me. And I think. How do you end up there? How are you able to get the exchange program there? So you just, you basically just apply and they do these like exchange things where then I think someone from Stanford like goes to code making for like a semester. How did you decide? Did you just find on the website and say, hey, I want to do this?

38:01Like how did that happen? No, so this was like after I got very interested in startups and I'd spent a couple of years there being very immersed in building, not as a founder, but as like working in startups and doing my own projects. So what were those projects? Tell me about like what were those projects you were doing? So the first stuff I did, that predated university a little bit, but I think like one of the things I'm kind of good at is like seeing around corners and like arbitration, right? I'm not a great developer. I'm not like the person's going to invent a new technology that does something.

38:27But back when I was in my late teens, Shopify, Magento, e-commerce has started to become, you know, commoditized in a sense, but it wasn't very difficult to set up a web shop in the same way it was like five years before. So I would go to all the local businesses in my area. There's one of them I think is actually still alive, which is a tennis shop in the city I grew up. And I'd be like, hey, what do you want a web shop? they'll tell me we would love to have a web shop but you know it's gonna cost me twenty thousand dollars like i have all these agencies pitching me and i'd go like i'll do the 500 you know 17 years old and 500 like rather than sitting in the being a cashier for like 20 hours right i'll go home set up like a shopify store i'm again to something like that change the templates upload the logo put the products in very simple stuff i took like a couple of days to do something great they would be really happy i'd be really happy so that's the first thing i did then i started like running their digital marketing so setting up like adwords facebook campaigns back when that was you know a very new product and it was like a lot of alpha and just doing that you didn't have to be amazing at like optimizing these things to just get a result out of it and that was kind of fun but also felt a bit like you know i don't want to like run a digital agency that sells tennis equipment for like a local shop in Denmark that wasn't really the scale of the ambition so like let me start to do some of my own projects um and then I did a few like affiliate things I had a I had a website with like protein powder so like just like back we would game the the you still do that today but SEO essentially again kind of early on in SEO it was a lot easier than it is today to get to like the the top ranking page on Google if someone search for protein powder.

40:10That didn't work very well, didn't make a lot of money on it. But at that point in time, it was kind of like a fun thing, right? Then I started the early days of Facebook. Once Facebook put the news feed out, there was all this, I mean, there still is today, but like the early days, I guess, like viral content. Do you remember like when, when these kind of listicles and like... Yeah, Facebook's pages were like a big thing. Facebook pages was exactly a big thing. That's how I started to build a Facebook page. And you put all this like, honestly crap content, right? Like 12 reasons you shouldn't go to the toilet after watching a movie or whatever.

40:44All this like random stuff. And I would just like sit and write these articles myself. I'd put them on a website and you would put them on the Facebook page and then you would have some kind of incentive for people to click on from the Facebook post to your actual website. So basically the BuzzFeed model, right? And then you would put like ads on the websites and you could make some money. I didn't, I was not very like successful at it, but I still at that point in time made maybe like$500 ,000 a month. for a little work. 500 ,000? No, no, 500 to 1 ,000 dollars. Oh, 500 to 1 ,000. I'm like, wow, that was...

41:13500 ,000 would have been pretty good. So I had like, I had like a few other small things like this that ultimately just taught me a whole bunch of like, how do things actually work? And I think that's, you actually said in the beginning where we sat down, right? We're like, you want to talk about how things actually work. And I think it's like, it's such a frustrating, it frustrates me so much when people want to talk about like strategy, MBAs, we read all these like business books of like network effects, all this stuff, right? Certainly interesting. But what's actually interesting is like, what were the UX things you did that drove some viral loop, right?

41:53What are the actual tactics? Those are the things that matter. But it feels like we put so much value on like sounding smart and strategic stuff. The frameworks. The frameworks. And I think when you realize, at least that's been my journey, especially as a founder, right? But even starting back then was like, all those frameworks, they make you sound very smart. They're great to know. Of course, I'm not saying they're not great to know. But what really matters is the execution, right? It's the tactics. And the tactics change all the time, right? So instead of talking about like, oh, Uber is a great business because of network effects.

42:22Like how many people have stood on like a content stage and talked about that? The real interesting thing is actually like, what are the 15 things Uber did to drive those network effects, right? That didn't happen randomly. That was someone who sat down and did all this stuff. Intentional. It was intentional, right? This is a result of like lots of experimentation or whatever. And I think what I learned back then was just, was a lot of that, right? You know, how do you set up a Facebook campaign? That's actually very useful if you eventually want to be not setting the Facebook campaign up yourself, but how other people do it.

42:50But knowing exactly what it does, what are the variables, how do you do targeting? How do you write like viral articles that actually, you know, people click on? So I've always been obsessed with the tactics part of it. I think in my early, before I started Syntisia, I spent a lot of time on doing this. After having done these kind of small projects, I basically worked in a Danish startup studio. So you're probably familiar with the model, but instead of investing in other companies, they raised the fund to essentially build companies themselves and then eventually hire external founders and to kind of run the companies.

43:23And I wasn't a founder, but I was basically second in command when we were trying to incubate these projects. and I kind of found myself really being most interested and really good at these kind of growth slash product roles, like figuring out like what are the viral loops to actually grow something, right? And kind of think about that. And I think that's where I kind of realized what I actually want to do is I want to build a company, right? I don't want to build like a viral block. I want to build something more kind of meaningful long-term. And do you feel that like if you look at your, Or, you know, I like this idea that every CEO has some majors and minors.

44:03Do you think like growth and viral is like one of your majors? Yeah, for sure. That's super interesting. I mean, if I could pick a few, that would be one I would definitely pick, you know, to have. How did you learn it? Like, how do you actually learn how to do it? I'm actually interested in it because I love the idea of being good at it, but I'm not. Well, I think it's one of those things where I just obsessed over it. So there was this growth hacking as a term that kind of emerged back then. Um, and I mean, first thing was that I found extremely interesting how, I mean, you can go like micro macro on this growth hacking thing, right?

44:37Or the micro, you have all these like small, like psychological hacks, right? You've changed the copy to be something that pushes people to click the button more or change the color of the button. Like these, all these like small things, right? Which kind of your logical brain tells you it's kind of stupid, but it really works because ultimately, you know, I'm a big believer in the fact that we're, we're mostly animals still. all the way up to just like when you build a product, like how do you think growth into that product, right? How's it going to be distributed? And especially with internet products, that was just very clearly back then and still today, such an important part of building a big, it's not just the best product that works, that wins, it's usually the best distribution that works, right?

45:12And so I think I just got obsessed with like reading blogs, reading books. There's a guy called Sean Ellis. I think he kind of coined the term growth hacking and he built like, again, like a message board community around this stuff where people would start like sharing like things they'd done that worked. I started building, you know, a good network of people and friends that were also obsessed about this same thing. And really just, you know, really like geeking out over it, I think was... Was there a version of this that exists today? I actually don't know. The probably, I feel like it's probably one of those things for like back then was very like guerrilla grassroots driven.

45:49People were like very ideological about it. Now there's probably a lot of these like frameworks and what is the Twitter guy, Nikita, guys like him really knows what's going on, right? But I don't know if there's a strong community as there used to be. My take on this is that, so the other reason, outside of finding it very interesting, right? I always think, when I think back in my career, it's kind of like I discovered Shopify and I was like, this is interesting. Most people don't know that Shopify exists or how easy it actually is to do. So I can arbitrage that, by going to all these like shop owners, setting up the business for them, right?

46:27My kind of entry point into working in this like style thing was actually SEO. It's because I'd done like the viral growth stuff. I kind of started to get a sense of SEO and I actually worked a corporate job for six months doing SEO. But I kind of felt like SEO is one thing for like, I'm the 20 year old, right? And there's all these people who've been doing SEO for 10 years and they're clearly much better than me. They have all the network, all the people, they're experts in this field. Why would you hire me? Why am I special? I'm not special. I'm trying to climb the ladder of becoming an SEO expert.

46:55And then I was like, there's this new thing. SaaS kind of is slowly evolving, becoming, you know, what we know of SaaS being today. There's this new discipline, which is all about like, you know, growth hacking. There are no experts. There's like this giant Ellis guy. There's a bunch of other people. Brian Balfour, I think, was one of the like, you know, main characters of that community back then. And I'm like, these people are interesting, right? but they haven't been doing it for 10 years. They've been doing it for like three or four years and they're sharing their learnings very openly. I can jump on this trend and I can be an expert.

47:29You know, instead of competing with like people who've been doing SEO for 10 years, I could be Victor who's thinking a lot about growth marketing. And that turned out to be a really great strategy for my career because you really stood out and you could do some, you had a skillset that very few other people had. But I think what happens with these things is that it starts with a very open community. People are sharing a lot. There's a lot of people who are passionate about it. But then it becomes slowly kind of corporatized, right? You start to like hide your tactics because all of a sudden, like you want to like keep the things that really work for yourself.

47:57And things become more like figured out. It's less like chaotic. It's less grassroots. It becomes more of like a discipline, like a formal discipline. And that's probably a good thing in many cases because then it's easy for other people to pick up, but it kind of loses a little bit of the magic, I think. Yeah. And it's probably a bit like with AI, right? Like back when I started in Syndisi in 2017 and AI was like, you know, definitely not very consensus. And you had like open AI doing lots of like open source that came out of that yesterday, I think, with a new model. But everything was like very open source, research papers.

48:30Everyone's like sharing everything all the time. But then you had like a GPT moment and all of a sudden all the big companies goes like, why the hell are we sharing all of our research and all of our open source models, right? You probably should be doing that. That's bad for business. and then you kind of get this contract and it becomes more like a discipline. And I think that's actually like, I think I feel like it happens in every time it's like most areas that emerge in some way, shape or form. And they're always more fun in the beginning, I think. This episode of HD and HD is brought to you by a brand that's close to my heart, Brex.

49:00When we started Brex, it wasn't just about creating another financial product. It was about solving the real gritty challenges that founders and startups face every day. Building something from the ground up is hard enough without having to navigate clunky legacy banks with endless fees. That's why Brex is the financial stack that scales with you, no matter where you are in your journey. From maximizing runway to earning yield on your cash, Brex was designed to make every dollar count so you can focus on what really matters, building your dream. So whether you're just getting started or you're ready to take on the next big thing, Brex has your back.

49:35Check it out and see why the world's most innovative startups like Anthropic, Scale AI, and Robinhood trust Brex. I definitely feel this, that in the beginning of Brex, I used to share all our tactics on Twitter and tell everyone, these days we keep the stuff close to the chest because, so especially of marketing, it's like, if everyone starts doing it, then it stops working, right? So you kind of want to be like in the bleeding edge of this stuff. 100%. I think that's very true. I think it's like the tactics you read about now are the tactics that worked like two years ago, right? and that's probably something to build on it.

50:09But especially with growth hacking and this stuff, people are really like guarding it. I remember like HubSpot was a big one where, you know, the whole like inbound marketing, leads, eBooks, webinars. I think it got to a point now that like, if I see an eBook on a webinar, I'm like, get away from it. You know, like I never want to download an eBook again in my life of like how much has been done. Exactly, exactly. It's so interesting, right? I think AI right now, it's like, we're talking about it. I think there's this hype machine thing of like, you know, it's, and I think it's, growth is generally good, but now we're in this interesting phase where it's like, basically just lie as much as you can, over-promise things that don't work, make a really cool demo video, go viral, and then hope you can turn that momentum into a product that actually works.

50:54That's kind of like, even that is maybe like kind of old news, but like I feel like last two years, that's been the way to start an AI company, right? Yeah. Just like say a lot of shit that's not true, make a cool demo video, shout and make it go viral and then hope that with all the momentum you're getting in you can actually like raise enough money to raise enough money to actually carve something out of that that's valuable and fucking awesome if you do we did a version of that ourselves as well but that's clearly been one of the best growth tactics and no one could have predicted that because the reason it works is because AI has this magical wow moment thingy where everyone is like captivated right the first time you use chat GPT your mind is blown the first time you saw a synthesizer video your mind is blown but that novelty effect is going to wear off and probably in three or four years, it's going to be hard to do the same thing because people just like saturated with AI stuff.

51:39They've seen AI videos. They've heard AI voices. They've seen AI clones, right? So you always want to be like ahead of the wave and the curve. I think the best growth people are really are the ones that are a few steps ahead, right? And you know, it's interesting. I think what I've been finding is that one of the hardest things to do this after you're already successful is that the people who are on the bleeding edge are usually really young. You know, they're like 18, 19, 21, 22. And then after you get older, you don't want, you don't know that many 22 year olds anymore. You know, all your friends are like kind of your age or older and, you know, the more successful or wealthy you are, like your friends are also successful and wealthy.

52:16And then like, they're also not 18, 19 and 20. Right. So I've been doing some effort to kind of get back to that, you know, like who are the like 18 year olds doing shit today that we should like spend time with. Man, but that hits so hard because I think about this all the time. Like, I feel like I'm becoming an old man. I can still talk about like my growth tactics from 10 years ago, right? It's so true. And I think, especially with this, I mean, what I just said with this AI hype thing, you see a lot of these like young 20s, early 20s founders, hacking away, living in like hack out, doing all this like cool shit, right?

52:47Totally. But it's like, how am I going to replicate? Like there's the, what are they called? Cluelly guys, I think. Which is like, they run it as like, I mean, I hope they build a big company on the back of it, but it's like amazing growth tactics just being like extremely provocative. cheat on everything lots of these like cool demos and it's just like that vibe works really well I think Lovable is also incredibly good at like and it's just it's just very hard for like you need that like youthful energy right and that like not giving a fuck not having anything to lose not being in the trenches of like all the people that that like really live and breathe this it truly is like a new generation right exactly like you know we I think we need to kind of as, you know, the millennials now, we need to be always thinking about, okay, how do we keep in touch with that generation and what's happening there, right?

53:41Because if we don't, we're going to kind of be, you know, I have this story, there's one of the founders of one of the largest private equity firms in New York. My assistant used to work for him. And she was like, you know, a core part of my job was I would like print out his emails and read it to him and he would dictate me the responses. and I'm like how would someone want to do that you know like it sounds so inefficient the reality is like he's had assistance since the like 70s you know so he kind of just doesn't never needed to learn and like my nightmare is being in that situation where like I you know I'm successful so I don't need to do something therefore I don't know how to do something and then I completely miss the boat you know and a new technology and a new trend and I think it goes so much back to that idea of like you have to be tactile right like it's like yeah even reading about stuff is like you really want to do stuff yourself you want to like be in them but you want to be there and like do it yourself um and i think one thing that that i had this realization every generation has their own like platforms in some sense right and i think the discord is like where stuff's happening now dude i feel old on discord you know i'm like i used to be like on And IRC and stuff like that, like back in the day.

54:54And Discord is basically actually just like IRC, just in a prettier interface, right? Totally. And also the forums are Reddit, right? Like Reddit is the new forums, right? Like the PHPV. 100%. 100%. But then it was like, well, you know, when I was like sitting in my teenage bedroom and I'd be like, thank Counter-Strike, being an IRC, and you're kind of always in this thing. But like now, you know, my job's like very different. I manage like a very big team of people, which is awesome, right? It's just a very different challenge. and you do get like pulled further and further away from sitting and like, and doing these things, right?

55:25A hundred percent. But I think I really realized just like, just like the Discord is like where all this stuff is happening. And now I like try to spend more and more time on Discord. And it is, it's just super cool. Even Twitter, I would say, Twitter is still X, I guess we're calling it now, right? X is still like a good place to get new stuff, but you really want like the alpha that's on Discord. So let's do a call out here. if you're a Gen Z and you are on some really cool Discord communities that are like really in the bleeding edge of stuff, like message us and we'll be super down to get into it.

55:58You know, like let us know where the cool Discord communities are. If you need some old people, let us in, please. I swear we can add value to your Discord community, like at least a little bit, you know, like. Oh, man. Yeah, man, I feel completely like I'm like, fuck, like I don't even know how to like, you know, use Discord. So I started using it recently and getting into some communities and stuff, but it's way more advanced, though, than the IRCs of our day. Because, you know, you can create, like, so many, like, automations in it, you know, where, like, it's pretty... It's almost like an operating system.

56:33It's actually a bit like Slack, right? I guess Slack's almost like the very corporate version of it. But Slack is only automatable by the Slack owner versus, like, I think Discord, you know, it's more, like, there's more automation that's done by the community. But it is like, I also, it's very interesting, like something like RSC back then, which was like such a nerdy thing. I remember, I love RSC. It's so fun. You would like squat channels. You don't, you remember you have to have like a bot in a channel to like claim it. And all these things, there's so many of these things that are basically ported over to these, like now, of course, discord and slack, whatever is like used by, I don't know, very, very large percentage of the world's population.

57:12But it's so cool how these things start early on, or you can trace the roots back already online. I think that's also, that's why. I think if you're an entrepreneur like you and I probably you try and look for those patterns all the time. I think that's actually the one thing I'm very good at is pattern matching and trends. I'm a generalist in the purest sense of the word, which I've battled a lot with actually, because I was like, what are you really, really good at? But I think that's the thing I'm uniquely good at. I think that's where you want to be on all these Discord channels. You want to connect the dots.

57:46I think that's super fun. And it's just on the internet, it's so readily available in front of you. Right. And do you have like a way to like execute on this a little bit more scale or no, you kind of just like try to do it yourself? Like, because part of the challenge I have sometimes is like, you know, I kind of notice these different trends, but I don't have time to go and look at all of them, you know, like. So for me, there are many different types of trends that matter. And I think there's of course the tech trends, right, which is we're very entrenched in that community. I care very deeply about cultural trends as well especially because I built a product which ultimately is about creating content right but also because on a personal level I'm very interested I think you know when we were with a bunch of founders a few weekends ago we talked about TikTok I don't remember if it was you or someone else but I was basically asked like how many people are on TikTok right I'm definitely on TikTok yeah there was a few other founders who were not on TikTok right I have this like very strong opinion like understand people don't want to like doom scroll you know like Like people don't like TikTok for all sorts of reasons, but I really think you should be on TikTok.

58:49Like that is where you understand the zeitgeist, the culture. People underestimate, I think, how much culture happens on like TikTok and social media today. 100%. Understanding like the trends, what's going on. I think I learned so much from just, even if it's like random crap sometimes. It's like being a VC is not being on Twitter. You can't not be on Twitter if you're a VC, right? 100%. I think you understand so much. So I would say operationalize it by like making sure I doom scroll half an hour every single day. But then, of course, I think reflect a lot upon like, what are you seeing, right?

59:21What is, for someone building a video product, TikTok is an incredible, interesting product to use and see what changes do they make, right? It's like, there's almost no video. Sorry, that's not true. But there's less and less text and interface, which I find very fascinating. Even when people respond to comments, they can do that video now. as you can just see like you can see these trends right and and and i think i think being immersed in that is really important depending on the company you're building of course so so let's go to the synthesia right like um how do you start a company tell us a little bit of the origin story so i um i had spent these like three or four years ish working in this like studio incubator in copenhagen and i knew i wanted to build a product um i knew once get out of denmark i knew didn't i wasn't super passionate about building uh you know business process tools accounting bookmarks that kind of stuff corporate cards exactly those kind of things i've always loved frontier technology i've always loved weird quirky fringe ideas um and i've always loved like science fiction right and so it's kind of like if i can marry those two things together that'd be awesome i was very i got very into crypto by the end of my tenure there and still to this day i is very interesting.

1:00:36Did you buy Bitcoin? I did buy Bitcoin. I actually did my Bitcoin. I bought Ethereum at like$4 or something like that. Oh, wow. But you sold or you kept? So, unfortunately, I've lost a bit. So, the Lunar crash, I remember that a couple years ago. That was a brutal one for me. Oh, no. I'm sorry, dude. I've traded a bit up and down. I still have done very well. But, yeah, I knew I wanted to build a company. By that time, I figured out that I think I knew what I was good at. I was good at like building products. That's definitely like the part of like building a business on the best. That's not like sales operations for sure.

1:01:11But building products, I think it was good. And so I decided to move away from Copenhagen. I moved to London. I would have moved to the US, but getting a visa was like very difficult. And I was young, didn't have any money, et cetera. And I didn't really want to take a corporate job. So I moved to the UK and I met this professor in machine learning. And we basically started like consulting together. this is just when the Oculus headset had came out uh when Palmar was I think 2016 and that was like the hype cycle of of 16 17 ish right and um I spent a lot of time doing VR projects we work for the UK government like a subsidiary there to figure out like how can we make the UK the best place in the world to build a VR company uh with some like corporate clients and just really immersed myself in that world.

1:01:59And I realized pretty quickly that as much as I liked the technology and I mean, still today, I think VR is like, it's amazingly cool. I feel like there's just like not a real market here. It feels like most of the things I'm involved in, most of the things I see, it's like demo stuff. People may pay for it, but it's demo stuff, right? Like there's very few people who actually use this stuff on a daily basis. One of the issues that I got interested in with VR was like outside the form factor, obviously being, you know, a huge hindrance for market adoption. The other thing was content creation because when you create something, a VR experience, you're basically creating like a lightweight computer game, right?

1:02:37It's about like 3D graphics, positioning, like it's all, it's like, it's very different than like filming a video. And so I got obsessed with this thing, like how do we make it easier to make content for VR? And one of the big ideas is a part of setting up this like massive Hollywood style capture stage in London called Dimension, which was essentially you go in imagine you have like 80 cameras in an array around you you go in you record yourself dancing or saying something whatever and you get kind of like a hot baked 3d asset out of that so instead of having to like sit and animate a character to do all sorts of things like how do you know that doing that would be a potential technical solution to creating vr content well that was i mean that was obviously like not just my idea right it was like by being immersed with people building vr experiences, researchers at like UCL and people thinking a lot about this stuff.

1:03:27And basically the idea back then was like, and I think it's going to be interesting to see what's going to happen with that. But the idea was basically, instead of like having to animate 3D assets and create an animate 3D assets, what if you can just record the world in 3D and import directly into a computer game? So we want you dancing a dance. If you have to sit down and painstakingly do that, like frame by frame, it's going to be a very tedious process, right? And very expensive. What if we could just have the studio, you go in, you make your dance, and basically with volumetric capture, and of course, computer vision, computer graphics, kind of like early ML stuff, we can essentially give you a 3D asset that you can just drop into a 3D scene, and now you could experience that in a VR headset.

1:04:07That would be a lot easier and a lot cheaper, right? In reality, of course, it's like very fickle technology, took a lot of time and money to make it work. But that was kind of like a big project I was involved in. and through that I met a lot of interesting people and among them I met Matthias Niesner who was a professor at the time he's still a professor and he had done this research paper called Face to Face he was also very interested in this idea of like content creation for VR but he did this paper called Face to Face which was not VR it was video and basically what it was was deepfake is is not really the word we use but I think for the sake of most people understanding it there was kind of the first deepfake technology the world saw essentially a neural network producing video that looked very, very photoreal.

1:04:53Hollywood has been doing that for many years, right? That's not because the idea of making synthetic video was like a new idea. Hollywood's been obsessed with this for like many, many years. But the big difference was that this was like a fully automated process that worked in real time actually, right? Did it start in Hollywood or did it start with porn? It started in research labs where like this paper, I don't know if that was actually the first paper, Whereas they're the first one that like worked well. I think it just started as like an interesting academic problem of like, how can we do animation with machine learning, essentially?

1:05:26Instead of like people have to sit down and do it. I think porn is probably like one of the areas where it's like used the most today. It's probably earlier. I remember first hearing about deep fakes as like scandals of like Hollywood celebrities that now were in porn websites with deep fakes and stuff like that. I mean, I think a lot of stuff early days was certainly a problem, but I think the research came much more from Hollywood, right? Because if you think about how expensive it is to make a Hollywood film, it's very expensive. A lot of that is visual effects. Marvel movies, like 100 million FX.

1:05:59Exactly, right. And creating 30 seconds of synthetic content with visual effects for a Hollywood film is like millions and millions of dollars. So there's a lot of research going into it. of it was actually also funded by like the visual effects studios, the film studios, like the more you can slice off that process and make it easier and faster, that's going to be, you know, positive of all those people using those technologies. That's kind of like where it started. And it's actually what the company started. So I saw this technology and the first time I saw it, I think it just had this kind of felt like I saw magic for the first time.

1:06:36And my mind really just started spinning and I just, I just kind of couldn't let it go. I think what, where it kind of clicked for me, I started writing down this thesis of like how I thought this was going to change the world. And it's basically like two big ideas. The first one was that with this kind of tech, the marginal price of creating really any content, but video, of course, in our case, it's going to drop to zero. Not just in like dollars, but also in time and skill required to actually do this. So just like, you know, you can, but you could back then, you can go into your bedroom and you can write a book and you can publish a book.

1:07:09You never have to actually leave your computer, right? Everything can be done in the box. that's going to be possible with video as well. You're going to basically be able to make a Hollywood film from your laptop, needing nothing else but just your imagination, right? And we're not there yet. This is in 2016. We're obviously not there yet, but I think that's going to happen in the next five years for sure, right? And that in itself, of course, changes everything about content and media, et cetera. And it's a huge economical opportunity for sure. So that excited me. That was the first part of the thesis.

1:07:39Dude, imagine Netflix's margins after this. absolutely crazy absolutely crazy right yeah but there's also there'll also be a lot more competition right so what but but Netflix at that end will be challenged right 300 million subscriber base or something like that right like it's it's going to be so interesting to see what's going to happen when this stuff like really works yeah the second the second part of the thesis was that what what always happens I've always been very interested in like media and culture and I think when you when you study the history of media right one thing that's very very obvious is that when we invent new technologies, especially new media technologies, we always end up inventing new media formats that are native to those technologies.

1:08:20So the idea of like user-generated content, for example, I mean, that wouldn't exist without a smartphone, right? Yeah. You'd never see, if we hadn't invented smartphones and everyone had a camera in their pocket, we wouldn't walk around with broadcast cameras and make TikTok content. Yeah. In music, right, you know, drum machines were invented back in the 70s, not because people wanted to make like electronic music, but because people didn't want to have to use a human drummer. So if you could use a machine to be the drummer for you instead, that'd be pretty good, right? Then what happened was people listened to these drum machines that are like, they're not very good.

1:08:50So we're not going to replace the drum. Maybe we will in some instances, but a lot of other people saw this thing and they picked it up and like, well, this is actually pretty interesting, right? And that turned into the basics of creating electronic music, right? That you have a drum machine that can actually play for you. A drum machine can play drum lines that a real drummer would never be able to play. a tempo is a real drummer but never be able to play synthesizers which was also an attempt to like can we synthesize the sound of like a piano or guitar or whatever which you can do with synthesizers but you can also create entirely new otherworldly sounds that you can create with a normal instrument you can sample so all this stuff is like new technology always really shapes the media formats that we consume and with this when we can generate video from code instead of having to run around in the physical world with a sensor clearly that's going to change video as a format as well, right?

1:09:39What all of us think of as video being today is essentially like heart-baked format. You record something. I mean, you play it back on YouTube. You play it from A to B. It's broadcast. Everybody watches exactly the same version of the video. We've done a lot of work and make it easier to make videos with like smartphone cameras and editing apps, et cetera. But video is still fundamentally the same thing because it still relies on you filming something with a physical sensor in the real world, right? and i mean we're gonna launch a product around this very soon but this was always the thing i was the most interested in more than they're just driving the marginal price of production to zero because this is where real big opportunity happens right new types of media formats and so if you think of what this will mean an example i often use is if you if you think of the difference between like a newspaper and a website you could argue that in many ways they're similar right it's a page of information or something and so the first websites they look like a newspaper on screen, right?

1:10:34Because that's kind of what we could imagine back then. But since then, we figured out that on a website, you get a lot of things you cannot do with a physical newspaper. You can update it in real time all the time, right? With one click, you can access like a gazillion pages on the internet. You can connect people in the comments field. You can add video. You can add audio. You can personalize the content to be different from you and from me. And obviously, today, the experience of like printing a million newspapers, sending them out to people around the world, and a million people visiting a website, it's like completely different.

1:11:02They almost have nothing to do with each other anymore, right? And that's going to happen to video as well. Video is going to become something that's personalized. Every single person that watches the video will have a different experience. It's probably going to be conversational now with all the, you know, LLM technology that's coming out. It's going to be generated just in time. And I think in five years time, what we think of as AI video, or maybe it won't even be called AI video, and a normal video is going to be completely different. Right now, it looks very similar. We're using new technologies to create old school media, but that's going to diverge very quickly.

1:11:32And that's opportunity that I think is much more interesting for multiple reasons. But that was basically the two kind of key ideas the company started with back then in 2017. Very interesting. And then tell me the, you know, what was kind of product number one that like first got product market fit or the first trials and then how it got product market fit and how did it evolve today? Because you started pre all this, right? 2017, this is like eight years ago. For sure. Yeah. I mean, we started back when the very first like generative AI that was like, you know, GANs was like a very early technology, like generate images of faces.

1:12:06I don't know if you remember that followed back then, but it's like, it's like super grainy, 64 by 64, like, you know, very different what it is today. We went out, right? And I went out to the market. I was in London at the time and I went out. I spent a year convincing this Matthias guy, I was 25 at the time, to come in to start this company with me. He was very against starting a company. He just wanted to like do research. And most people came to him was like, hey, hey, let's use the technology to create like funny Snapchat filters, something like that. We can sell it to Facebook. And it was always just like, I don't find that very interesting.

1:12:37So I think the reason he wanted to start a company with me eventually was because I was like, look, this is so much bigger than like doing Snapchat filters with dog ears, right? Like there's actually something very real, very transformative in this tech. He eventually said, okay, I called up Stefan, my other operational co-founder, we met in Denmark and said, hey, I have this crazy idea. Do you want to try it out with me? He's like, yes. We found one more professor from UCL in London, Lourdes Accapito. And then we went out and try and we found our first PhD student to build the early iteration of the product.

1:13:10And we could pay him because we'd made a little bit of money on crypto back in, that was like right in one of the first bull runs. We went out to the market and we tried to raise money. It was absolutely impossible. People thought, you know, I was not an AI researcher or professor. I was 25 years old. I have on paper, I think I had a lot of reason to be the right person to start the company. Looking back now, we talked about a lot of them earlier here. But back then they looked at me and like, who the fuck is this dude, right? This idea is absolutely crazy. Not interested, which is fair enough because it was a crazy idea back then.

1:13:39So we got turned down by basically plus 100 investors. Eventually we're just running out of money. My co-founder Steffen sent an email to Mark Cuban because we had like this long list of people in the Venn diagram, like media and technology. He responds back within five minutes. We have this like 13 hour email conversation with him. And essentially he ends up investing a million dollars without ever meeting us or talking with us on the phone. Oh, wow. Which was like definitely the craziest 24 hours in my life. And that became the initial spark of like Synthesia for what it did. Which valuation did he get into?

1:14:11Five million. Wow. So he crushed it in this investment. He crushed it. Wow. But I am eternally grateful for Mark. Mark believed in us when absolutely no one else did. And the reason he believed in us and my biggest takeaway from that was that every investor I went to was like sure, Dan. sounds cool, right? Hollywood films on laptops, cool idea, not investing. Mark had actually implemented the face-to-face paper himself at home with a tutor. He was completely convinced. We didn't have to talk about the thesis at all. He was 100 % agreement, like this is going to happen. He was only interested in who we were as people, how we were thinking about the business and product and strategy.

1:14:50And so my take on that was like, we could raise money from him because he already believed in the vision, right? Most people back then did not believe in the vision or like the kind of trend that we were investing behind. I think in general, like with investors, of course, now AI is like very consensus, but you always want to try and find investors that you don't have to convince of like, this change is going to happen in the world. I'm the right person to do it. If you have to convince them that this change is going to happen in the world, generally like... It's going to be hard. It's going to be really hard, right?

1:15:15Especially if you're very early. Anyway, Mark gave us a million dollars. We had to start a company, right? So we sat in a room and we were like, okay, we're definitely not going to be able to build technology that can make Hollywood for laptops for a million dollars within the next couple of years. That's like totally unrealistic. What could we potentially do that actually could work? And the first thing we did was this idea of like AI dubbing, which now has made a comeback because the technology is actually really work. But the idea was pretty simple. You know, we're not going to try and generate like the entire video.

1:15:44We're just going to focus on translating video content into different languages, but without the viewer being able to tell that it's been edited, right? So you take an advertisement shot in English and you want to, you know, take that to the Brazilian market. What we would do is we would then record a new voiceover. This is before text-to-speech was good enough to actually, you know, pass the Turing test or, like, not be really degraded quality of the content. And then we would take that and we would reanimate the face so that it looks like you were speaking in Brazilian. It worked if you looked straight at the camera without looking away.

1:16:18Doing a 30-second clip was, like, two PhDs for 10 days, like, struggling to make every little pixel, like, fit perfectly. So it was not like a very good business at all. We did like a million of dollars in revenue in like two or three years, which was like, okay, and kept us afloat. But it was very clear that this was not the path to build a big company. You know, we'd end up becoming like a visual effects studio with a proprietary technology. And that was not what we set out to do, right? And so what happened was we slowly realized that we're trying to sell this to video professionals, people that work with video, the most obvious people to buy this technology, right?

1:16:54is obviously going to be the Hollywood studios, the video production agencies, advertisement agencies, all these people that work with video all the time. We try to build a product for them. I since have learned, and I think for like most like truly great companies and great products, it's rarely the most obvious buyer who's the first to adopt it. Actually, it's often the last one to adopt it, which is also why I think something like Hollywood would be some of the last to adopt AI video technology because I had this experience. These people's house were not on fire. They thought what we did was like cool, It's a total vitamin, not a painkiller.

1:17:26They already get to be with lots of awesome video content. You know, we had zero control over the go-to-market process because... Innovator's dilemma, right? Like, you want to go who don't have anything, right? Exactly. And that was a lesson we learned. It was like, because we were just, I think what we did right back then was like, we're just obsessed with learning and understanding like video from first principles. So we were, me and my co-founder, Stefan, who are not quasi-technical, not at all building the AI and alternative AI, right? More like a product person. He's more like a finance operations.

1:17:59And I think a lot of VCs hated the fact that I wasn't an AI researcher starting the company. But I think that actually ended up being anything that saved the company because we were just obsessed over just talking to everyone we could about video. And we did this very famous campaign with David Beckham, where we made him speak nine different languages, which was one of those things like from the outside world, I was like, oh my God, this company is a huge success. did it like David Beckham thing. In reality, I think we got paid like$1 ,500 or something like that for like four weeks of work and all this like craziness.

1:18:26But that gave us a lot of inbound and we managed to speak to a lot of people. And essentially, what we figured out was that what's more interesting than all these people who are already making videos, all the like millions, if not billions of people in the world that are desperate to make video, but are not making video today. Because they would contact us and be like, hey, can you help me do X, Y, Z? Like, I really want to make a video, but I don't know how to use a camera. I can't get the budget from my boss, whatever. And the more we spoke to these people, the more we realized that these people's house were on fire.

1:18:54They had a very real problem. They wanted to communicate something to someone. And they knew that if they sent them a 20-page PDF document, no one would read or consume it or remember any of that information. So they wanted to do it in video, right? But they had no way of actually doing that. In the corporate world, like a lot of these people was telling us that, you know, what we really want is just like a talking head style video. Like they're not asking us to create AI videos of people like flying helicopters and jumping out in parachutes, right? They just wanted the video like of someone talking to the camera, and then they would probably cut it in with a real video of someone assembling a product line, whatever thing that they wanted to do.

1:19:29And that was a much, much easier problem, of course, than saying, we'll build AI video that can do anything. So we, and another thing was that these people, right, when they judge the quality of the videos we can produce, they're not going to judge it against a real video. They're going to judge it against a 20-page PDF document, which is their real alternative, right? That led us down to this idea of making avatars, which is the kind of core technology we developed and were the first in the world to launch. We started developing that in probably 2019, I think. In 2019, we launched an MVP where we could look kind of like a SaaS platform.

1:20:08You could log in. You could pick on these avatars. you would write a script just like you do today you could hit make video but it was a preset avatar it wasn't yourself yet this was purely just like form submission there's actually no tech behind it but we just wanted to test out like how people react if we give them this thing where you just like select the avatar put in some text and get a video back so what actually happened was when we clicked that button we would get an email in the office and we would like panic to try and make this video because nothing was automated and who was actually doing the videos at this point so we had like a team of like video editors sitting and doing it.

1:20:40We had to like record a voiceover. It was like a very, very laborious process to create like a single video. What was the tech at that point? So the tech was the same as the dubbing. We would have a real video of a real person, but instead of translating into a different language, we just replaced the like spoken line. Does that make sense? And so that was basically what we would do, right? But it was like a very intense process to make it work. But the thing we discovered was that the magical moment for someone to go to a website, put some text in a box and receive a video back, even if it took 48 hours, something like that for us to actually do it.

1:21:13It was amazing. It was magical. People were blown away, right? And that gave us the confidence, okay, let's figure out a way to actually scale this. And by 2020, we launched the first AI avatar product to the world, which actually worked. It was like, you know, go to the website. It's full machine learning now. Full machine learning now. Go to a website, pick an avatar. Of course, like very, very basic compared to the product of what it is today. Put in some text and we'll generate the video for you. And back then, I think it took like 25 minutes, something like that. The voice was not very good.

1:21:42This was before text-to-speech really kind of broke through and became very, very, very good as it is today. So it kind of sounded like a car GPS. The facial animations were not like that great either. But it was just, it instantly went viral. It was like the Google Translate kind of voice, that kind of vibe. Exactly. Yeah, exactly. But it instantly went viral. Started getting all these people coming into Synthesia. You know, this was like, at this point it was not even a product. It was like a demo page kind of thing, right? All these people come to us and we would start to look at all these people and sift through, okay, who's doing this because it's like cool and novel and they want to show their friends and their colleagues and their mom and their dad and they're making a birthday video with someone and all that stuff, which is great for us because that keeps the viral loop going, but that's not a real use case.

1:22:27And who's actually coming back, right? Who's actually using this for something real, if any? We were honestly, when we launched the product, we were like, we have zero doubt that this is cool. The question is, is it useful for someone, right? But it turned out it was useful for someone. You know, we started, one of the early ones, remember, is like there's like this really big fast food chain and they're making videos with this demo thing. And I called him up and I'm like, hey, you know, what are you using these videos for? You keep coming back, right? Why are you doing this? I didn't say that, but I'm like thinking because the quality is not very good, right?

1:23:01This person tells me, well, look, I have to train millions of people to go work in restaurants every single year. And the way I do that today is I send them a 20 page, 30 page, whatever PDF document. No one reads it. And even if they do read it, nobody remembers any of this. We've tried doing a bit of video and that works really, really well. People retain so much more of the information. But the problem is we operate in 50 different countries. Policies change all the time. Processes change all the time. so every time we've made a video it's out of date before it's even published so which fast food was this can you share I probably shouldn't share the name but it's one of the world's biggest fast food chains okay and I was like you're a genius I would like very respect for that you know mad respect mad respect but you know we just had so many people experimenting on that and we found several other people this way and kind of it makes a lot of sense the thing we realized was that when everybody thinks and it was so hard for investors also to understand it even though we had an easy time raising money after the company started working, so hard for most investors to wrap their head around because people would see the videos and they were like, again, it's clearly cool, but who's watching this, right?

1:24:12Because if you and I, most people sit down, you think about like video, right? Close your eyes, think of video. What you see, you see like the latest Netflix series or you saw some like really well-produced YouTube thing that someone did, right? You think about all the video content that bubbles to the top of the creation, right? The reality is that most of the world's video content is pretty shitty. It's pretty boring. It explains some like process. It's like, you know, it's not like a great video necessarily, but it serves some purpose for someone. Even if it's a video that has a thousand views, but it's a thousand really important views for some business, that has value, right?

1:24:49And so what we discovered was that our market is actually all the like worst videos in the world. It's all the text documentation in the world. It's all the slide decks in the world. we're not at all competing on any use cases that people are already making video for today because of like good videos, right? Because people are already making good videos the value proposition just doesn't work because they're a little bit more efficient but it's not transformational. Exactly. But no, actually not even that because the quality is just not good enough. Like if you do a marketing video with an avatar back in 2020 or 2021 like that marketing video is not going to perform very well because the avatar is going to sound like a you know, Google Translate voice it's going to look pretty janky and that's just not that's not what you want to like show the world about your really like cool marketing product right so I think that was a very big unlock for us we figured out like if you really think about all the world's video and all the world's information that's being communicated you know that's the target market for us that's where this technology really works it's not the sexy use cases it's not the Hollywood films it's not like the cool ads that you see when you're browsing on YouTube whatever it's a lot of like very practical utilitarian content and then we started building for that and what we learned early on was that the people making a lot of this content were not like video editors necessarily.

1:26:02A lot of them were PowerPoint users. And so instead of just having this technology where you had like the first iteration, this was like an avatar clip generator, I usually say. So you would pick an avatar, but you would get out of that was basically a video of an avatar talking to the camera. Most videos are not just a personal avatar talking to the camera. You have some animations, you have some text on the screen, you cut to like a screen recording, like you put some other like media in there. And a lot of these users was like, well, it's great that we can make the avatar thing, but I am not comfortable using Adobe Premiere.

1:26:35So it's very hard to finalize the videos. And so we started building this tool around it, which is, I'd explain it like, I did it very purposely when designed. It's like, I want PowerPoint users to feel at home within 30 seconds. Yeah. Kind of the editing experience. Editing experience. We built the editing experience and it's kind of a mix match of like Canva PowerPoint style, but made for like a group of people that never have made video before. That was really successful. And today it's amazing. It's still lots of work to do, but it's a great editor where you can really make like really great videos and you can pick it up very quickly if you know how to use PowerPoint, which most people do.

1:27:11Man, you know, it's so great. I know you don't like frameworks, but this is like so like textbook innovator's dilemma where you get something that, you know, it was not good enough for the mainstream market, which is like Hollywood right and then you start with like a use case that you know they don't have as much demands because you know it's they want something that's easier and cheaper than with the alternative and then you know years go by by the time you get compete to the follywood guys the thing is like so good you know and so much better you know and so this is like how innovation happens and we figured out very early on that the rate limiter for the tam was based the quality of the device in the avatars right and the more that's gone up over time the more we've unlocked more and more use cases that are more you know higher value use cases more external facing use cases etc but if you look at like i think it's always interesting you look back at like the last whatever like nine years like all the competitors that we've had throughout the years and like why have all of them or most of them gone bankrupt it's all been because they tried to tackle use cases that were just not mature enough yet the tech wasn't there yet tech wasn't there it's like every yes everybody wants to make like personalized sales videos or personalized marketing videos and it's for sure going to be a thing the reality is just that the tech's not good enough yet right and so what happens is you get this kind of fake signal where you build you build the product you get lots of interest everyone i love it people on twitter are like oh my god this is like the most amazing thing you know this is going to be fantastic and you go oh my god people want the product it's amazing but then you can't deliver on it and then eventually the rob is going to hit the road right like you can't keep like raising money and over promising um forever at some point rob is gonna hit the road and so i think what we did well is we focused on the very like boring utility driven use cases we built on top of that and now we're getting to the point of the market and technologies where they are actually starting to become useful for making also more like interesting entertainment the advertisement like content and and and of course you know i think we We have a great hand because we've actually built a really big business that's not driven by people making demos.

1:29:19It's driven by people having real utility. But it's very hard because it's so captivating to chase the cool stuff and the fun stuff and the hype-y stuff. I think that the VCs, I mean, VCs, right? No VCs like the fact that we were doing a lot of training videos three or four years ago. It's like, oh, it's like small time. Don't really like it. These teams don't have a lot of budget. And all those things are actually true, right? But there are so many things pointing to this being the right strategy. One of them being what we just discussed. The other one, which I actually also find very underappreciated, I think we discovered this early on, was if we had made a product for making marketing videos early on, marketing people, I mean, and still today, right?

1:30:02CROs, CMOs, whatever. The amount of like products and cool products that are pitched to them every single day is absolutely insane. I mean, you know, there's a Brexit, it's hyper, hyper competitive market. Lots of products out there, very, very difficult to like break through the noise. And these people are evaluating tons of like super cool products all the time. We learned early on that what's very interesting is if you go to these training teams, nobody gives a shit about them, to be honest, right? They're not pitched very cool products. They pitch like a new, boring, dusty LMS system like once in a while.

1:30:34And we built this product for them, we targeted for them, and they loved it. super easy to get in touch with them, right? They're all just like, this is so cool. This makes my job so much more fun on a personal level. I'm creating so much more value in the company. And so it was one of those beautiful things a bit as we talked about, but like growth hacking early on, right? Where we realized that not just is this a great product for these people, this is the most amazing distribution channel. Because what we discovered later on was that it's not just easy to sell to them. Who do these people create content for?

1:31:05The rest of the company, right? Right. And so it's the perfect Trojan horse to get in with these teams that create content for a lot of internal use cases because everyone starts seeing all this content. And then what has happened now over the last two, three, four years, whatever, right, is that as those videos have become better and better and better and better and better. Now the product marketing team is like, hey, this stuff's actually pretty cool. Why don't we make like 40 videos about like how our product is better than competition and something like that. Right. And then we do the classic land and expand.

1:31:34And today, you know, we work with like, you know, so many teams in the enterprise across so many like different use cases. But it's one of those things where like it was very non-obvious from the outside. A lot of, again, a lot of VCs like definitely did not like that go-to-market. But I think it turned out with some, definitely some luck, but also I think a little bit of skill and being very like driven by this like idea of like providing real utility, not just demos. We managed to like, to really, yeah, I think capture the market in a really big way. Dude, I'm very sold. I kind of like almost clicking the buy button right now.

1:32:06But like, so what is the core tech today? Because I imagine the models themselves, you know, it's hard to compete with the labs, right? Like, what is it that you guys decide, hey, this is proprietary tech? And what is it that you're like, look, whatever the best model is, we're going to do? I mean, look, this has been one of the hardest things of building this company over the last eight or nine years, right? It's like, well, maybe not the last eight or nine years. Let's say the last three or four when model development really started accelerating. There's always this thing of like, what do we build ourselves?

1:32:35Where do we think we can actually compete in a market where progress goes like this? When do you plateau and start optimizing your model to be fast and low cost? And when do you keep just like jumping to like the next shiny new thing? This stuff is really difficult. And I don't think we've done it perfectly, but I think we've done it pretty well. So there's a few things. I think the first is really being very intentional about what kind of company you're building. I think we started off being... I mean, the foundation model was not the word that was used back then, but we're definitely developing this tech in-house ourselves.

1:33:13But that was out of necessity. There was no option than building it ourselves. And I think now what we've switched to is being much more like we're like an application layer company. Everything I do, I do by working backwards from like a workflow that my customer has that I want to make perfect. And that starts at the model layer where we essentially replacing the camera. But it goes into like the actual editing part. It goes into the collaboration platform around it. It goes into the publishing platform where we have our own video player you can use for analytics and insights, enterprise, all this thing, right?

1:33:43So you work backwards from that. And essentially, how do we build the best product? And we can build the best product by integrating third-party models. We'll do that. which is launch to be free, for example, in the product, complimentary to our own. But when there isn't something available and our customers need something, we should build it ourselves if you think it makes sense. So in the space of AI video, for example, there's all these generalizable models that do, you put anything into these models and they'll spit it out, right? Like give me a dog skateboarding on the beach, give me whatever you can think about.

1:34:14And that's super cool. It has tons of use cases, but they also have a bunch of downsides, which, again, if you're not tactile with these things, it's hard to understand. A lot of these things for many years couldn't produce speech, for example. That's a problem in our use case, right? They also are very bad at identity preservation. Our customers want the avatar to look exactly the same every single time. Our customers are also not creatives, right? They don't want to spin the AI model slot machine 20 times to get the right prompt to get what they want. They want to put in some text and they want to get a video out that works.

1:34:46Consistency. Consistency, right? Yeah. they want it to be controllable. Like, and these details matter a whole bunch, right? And so we've built our own appetite models that optimize for what our customers actually need. And now we like blend them in with other models that can do these kind of more generalizable things. But it's something we are constantly evaluating for sure, right? Because it's very obvious also that, you know, compute's going to matter a lot. I do think we can win with a team and the compute cluster we have for years to come, definitely on the right mix of model, right? So the cost, latency, identity preservation, controllability, all these things that matter if you create lots of videos every single day.

1:35:28But that also means that we're not building the right models if you want to build like entertainment, movies, Super Bowl ads, that sort of stuff, right? And I think the other thing that I think constantly is interesting is this idea that when you are in space like this, but the progress is advancing so rapidly, it's so hard, right? Do you try and like ride the wave up? And you probably do. But in video, you know, just over the last three years, there's been like three paradigm shifts of like different types of technologies, different ways you could build it, right? Is it like diffusion? Is it like token-based?

1:36:02Is it... And having to pick a lane is very hard. And the talent war too is insane. The talent war is insane, right? But then you build something and by the time you're finalized building something, you've all agreed that, okay, this is probably like the right, this is the way this problem is going to get solved. Then we get very close to launching it. Someone comes up with like a new way of doing things and everyone is like, okay, this is clearly probably the way it's going to be solved. What do we do with this tech we have now, right? Do I spend six months putting into production and getting out to our customers?

1:36:30Or do we say, you know what, screw it. Let's just go for like the next shiny thing. And that is like very, very difficult to think. Now I think what we're starting to see is like in the tech space, obviously like there's like some level of saturation where I get marginal improvements. I think that makes it easier to say, okay, let's take models, let's tune them for like our specific use cases and so on when there's like some level of saturation in the market. When the market is still progressing like this, it's very hard, I think, to make the right calls. But for me, I think we increasingly think of like the models as being, when I think of like what are the most valuable parts of Synthesia, when you look at it as like an asset, I don't think it's the AI models, right?

1:37:06I think that's actually going to go down over time because this stuff's going to get commoditized. The real mode and value that you build It's your customers. It's the workflows. It's, you know, the editor that we have, which helps you put together the right video at the right time. We have like a co-pilot that helps you do all this stuff. So I really think the model development is essentially like a go-to-market advantage, right? If we're like six months earlier to market with like the highest quality avatars and a competitor, that matters quite a lot. And that's probably worth quite a lot of money, right?

1:37:33And what do you think, like, if you're like 18, you know, on Discord, and you want to go into this creative world of video and everything. What do you think is the skill that is going to be valuable in the future, right? Because before it was, you know, knowing how to use Premiere, good of cameras, light, you know, like kind of the traditional kind of Hollywood movie making stuff. What do you think is going to be look like in the future? I think it's got to be taste. And I think that's going to be the case in like most disciplines. Taste is one of those things, like what is taste, right? I think taste is a combination of like deep understanding of like the cultural zeitgeist, like what stands out, what gets people's attention, what are people talking about.

1:38:16Storytelling, which is, you know, you have some methods that you want to put out. How do you tell a great story that's like captivating? Curation of like ideas and different ways of doing things. I think those are going to end up being the most important things, right? If you take writing, I always go back to analogies to like writing because I mean, we don't think of writing as like a creative technology really, but it kind of is, right? Writing? Like writing things. text, right? Like writing today is like operating word is not the skill, right? Back in the day, there was a time where I was actually like operating a printing press, like printing stuff was like understanding that was like, was an important part of like making text content or books or whatever, right?

1:38:54I think we're going to get to the same thing where like the actual production of the final content, that's going to be completely commoditized. It's going to be very easy to do. And so it's going to be, again, it's going to be about taste, right? How can you create something that's culturally relevant, serve the needs of the business that you're working for or if it's entertaining people or whatever. I think that will remain the most important thing. And then I think as we get into this world where increasing video becomes programmable and becomes less about just broadcasting content, becomes more interactive and conversational, what does that mean, right?

1:39:27That's an entirely new skill. I don't know it. You don't know it. Soon we're going to have the tools to figure out if you can talk to a video, how are you going to do that, right? is it gonna be after you've done watched the video, you can just ask it a question. Is it gonna be midway in the video, you can stop the video and ask it a question. Is it gonna be the video looks for your camera to see if you're bored? And if you're bored, maybe changes up the content a little bit or personalize it. Like there's gonna be so many new things and ideas like figure out about this like new media format. And that's probably something where I think that's where we need like young people, take this stuff, figure out where it's valuable.

1:40:02My favorite example of this is Twitch. It's like when we, I think WebRTC probably was the technology that enabled real-time streaming in the browser, right? Everyone can basically live stream from anywhere, which used to be, you need to be like a broadcaster to do that. I'm sure all the McKinsey consultants back in the day would have said, oh, you know, we do this, we'll see like so many new TV stations pop up, like people will do like live TV and cooking shows, whatever. Definitely no one had predicted that the billion dollar use case at first was like watching other people play computer games.

1:40:31Yeah. But that was discovered because the guys at Twitch were like experimenting with like a bunch of different things. They did this like reality TV show, I think, to begin with, but it would go around and film. But they figured, well, the problem with that is that if you're like jumping into the stream midway, you have no clue what's going on. Then they started like filming people gaming and probably also like part luck, part skill. They figured out, well, gaming is actually great, right? Because if you already play Counter-Strike, you can join the stream at any time. You have full context of what's going on.

1:40:59Yeah. Plus people are, you know, you know, very online, usually if you're a gamer and then they figured out, well, a live stream is not just a video. It needs to be like a comment field. It's like you can give people tokens, like all the stuff that makes Twitch and like live streaming a thing. And I think we're going to go through the same thing with AI video where we have to figure out what are these new formats. And I think it's going to be much, much more transformational than like the switch from like static YouTube to live streaming. I think it's going to be 100 or 10, 100 X in terms of like how different it's going to be.

1:41:28and I think that's as a young filmmaker or like young video person I think that's as this stuff becomes available that's what I would spend my time on What about right now? What is the skill that's required right now? Is it prompting? Is it kind of stitching together these models? Like what is the skill that you need? I don't really think like prompting also I think it's underestimated how hard prompting actually is I think we all make it sound very easy like just type what you want but it turns out that I have a lot of friends that are artists and they're like 100 times better at prompting than I am right because if you have something in your head and you want to bring it out as a prompt, that's a skill, right?

1:42:00So I think definitely like prompting video models is probably like the one tactical skill. That's very different from what would have been like five years ago that you should learn. Makes sense, makes sense. To close it out, right? Like, I guess as you, you're 34 now? 33. 33. As you kind of reflect, right, on your journey over the last 10 years and you had to reflect on like a few things that you were like, look, these were the things that, you know, got me here and the strengths. And these are the things that, hey, maybe they got me somewhere, but now I'm going to have to change in the future.

1:42:43Like, what would be those things reflecting on your journey? I think the thing that got me ultimately is very much my like curiosity and my deep interest in like how things and systems work and how people work and how culture moves. Because I think what I have, what we have done well at Zendizia is we've kind of seen around the corner. We believed in something, even when a lot of people did not believe us. And we managed to stay alive long enough to see that come to fruition and sequence a company in the right stages, right? The whole journey of like, one thing is like, I've heard there's a lot of people where you can like say something's going to happen in 10 years that's not really the hard part of it right the hard part is actually building a company and sequencing it along all the different chapters of like different stages of technology so i think that's i think that's one thing that i've done well i think the other thing is i think i i always i really have this like deep city thing in me as i said earlier where i just generally people say tell me something I assume they're wrong, not right.

1:43:47I don't know why I'm wired like that, but I just am. And I think that's, in many ways, can be annoying if you're around me. But I think it's actually really a strength. I really truly believe that we have the tendency to think that we live at the end of history, right? And that all the kind of moral beliefs we have today are going to stay like this forever. That all the technology views today are going to stay like this forever. And it's just, if you look back at any point in history, like let's go 100 years back, right? Most of the things people believed where you were wrong or has changed significantly and definitely the world around them has changed the physical world and the technologies that we use, right?

1:44:22I think having like that kind of mindset all the time I think is really valuable as an entrepreneur. In terms of what I'll need to change, I think, I do think it goes a bit back to the discussion we had about like Discord and, you know, staying in the know. I think when you're, when you become like a, more like a CEO, CEO, right? Like when you're like, you know, the hundreds of people and your job essentially does become all to be kind of like a manager and a diplomat and a politician to some extent. I think really making sure that you spend time on like going back into really touching like the world of like new and emerging and fun things just like, you know, I did like 10, 15 years ago.

1:45:03That's something I think about how do I operationalize that in a way where it's not just like me remembering to spend some time on the likes of getting caught in a rabbit hole every other Sunday, right? Like, how do you really like stay very close? That is something I think a lot about. It's awesome, man. Those are great reflections. And I'm sure the next decade will be even more exciting than the last one. Thank you so much. Thanks, Adam. Thanks to our friends at Atomic Growth for helping with production and distribution.