AI That Works for Customers: Synthesia CEO Victor Riparbelli Explains

well Victor thanks for sitting down to chat with me we met I think it was a year and a half ago uh at this event that Excel was doing uh these shared investors that we have and I've always just been really impressed with you as a founder and with what you guys are doing at Synthesia so can you give a quick elevator pitch on Synthesia the product what it can do yeah sure and thanks for having me I think cth we the world's Premier AI video platform and at its core we help our customers communicate their messages the most efficient and engaging way possible and it's pretty clear I think to most of us that in 2024 that's not text or slidex that's video and audio so we provide one platform where you can create videos without having to use cameras so for the creation step we have all of our AI models AI Avatar is kind of like the technology I think that we're most well known for we have ai devices we have full video editor um that's all packed in a nice modern collaborative platform so can write with your colleagues work with them leave comments we can share the content with your audience so we provide of course like a video player but also a video player that's been optimized specifically to work with AI generated videos so for example our Video Player supports multiple languages um and a bunch of other things and we work with companies from like all sizes we work with some of the world's biggest companies um smallest companies work with individual creators but really the core of everything we do is we help you deliver your message in the most engaging format possible and you guys started the company was it 2017 or 2018 2017 I actually went uh this morning back onto the way back machine on internet archives have you ever done that before and uh what I think is amazing is like your guys uh hero copy was like pretty much the same back in 2018 and I was surprised because I would have expected maybe you started in a different direction you kind of found your way to the Avatar generation the video generation but it seems like that has always been the vision from when you started the company even what is it seven eight years ago now in some senses yes I think the vision hasn't changed I think um we had to figure out like how to sequence our way to get to where we are today and where we want to go in the future right so it really started um I mean we did what I think most VCS will tell you not to do is that we fell in love with the technology back in 17 when I met my co-founder mati nner uh who done the research paper called face to face which we kind of INSP Spirit at least originated from we both got excited about was that even though the technology that point was like very immature it's very early we felt like in 10 years you're going to be able to make a Hollywood film from your laptop without needing anything else in your imagination which in one hand is like a kind of a crazy idea I think to most people on the other hand like if you take the written word everyone today can with their computer write a book and publish it without need more than the imagination and we don't think of that as something special today because we all grew up in the age of computers but that is actually very special in you know the history of the world as Al felt like this is going to be the next big thing it's going to change everything we know about media production and that idea was kind of thr Fring to us that we felt like okay let's try and start a company to build kind of like towards that uh not necessarily because we wanted to make like Hollywood films but like the technolog is going to be so powerful going to be able to do these things right and back in I forgot what year it was but a couple years before we found the company Sony got hacked and all their internal emails got kind of put on the internet and as we were raising the round we had this like list of funds of course and every single fun

turned us down generative AI wasn't really that hot back then H then we had a list of like you know business angels individuals that we felt would like potentially be interested and Mark fit pretty well in that V diagram of like media and Technology yeah but getting in touch with him is obviously difficult and we were sitting like in London have no connection to him but then we downloaded this hack because my co-founder Stephan thought maybe his eil email was going to be in there and his email wasn't there and then we sent him an email um and he replied back within 4 minutes saying that uh he thought this was pretty interesting sent us some questions and we replied back to those questions and he sent some more questions and we did this for 12 13 hours I think oh my God uh until like 4:00 a. m. in the morning uh UK time where he said that he'd put a million dollars in given that diligence check through so we never went on the phone with him and we never talked live and I think my the lesson that we learned there I think is that we were basically going out to the market and we were saying hey this crazy thing is going to happen like you know with div vision and the thesis essentially yeah and we had to start from scratch like everyone was like it sounded bat crazy at the time um it wasn't like the current thing I think it is like when you're so much maybe ahead of like the market you really need to find those like missionaries who already share your view of the world right and they are out there it's just about finding them and for us it definitely turned out to not be like London based VC firms where everyone who worked there like used to do private Equity right like that's not the right type of investor for that type of play so right so

fast forward to I guess a few years do you feel like there was an inflection point with the either the product or the market or the company like a single inflection point you can kind of point to because I mean correct me if I'm wrong but I don't think you guys were on this like linear growth story from 2017 to now you know just from our conversations like it's really exploded over the last two years can you point to one or two things that you feel like changed that got you on this track you're on now yeah so I mean the big unlock was in 2020 basically when we got the first round of funding right we sat in room and we're like okay this round was not very easy to close we probably should treat this million dollars as like the only Lifeline we have to prove that we can get to the next stage and so as much as we thought it was very cool with like you know this Hollywood films and laptops and like that whole Vision we're also you know very realistic that I mean the Technologies are nowhere near there it's not even worth starting to build that so what can be the first step towards that Vision that we think we can feasibly both build and sell right and so the first thing we kind of began building was basically AI dobing which was the idea of like not like we do today predominantly like generate entirely new video but take your existing video shot with a camera and translate that to a different language the way we do that is that we take that video we'd replace device over and we'd reanimate the facial expressions to fit that new language and we managed to build that and we took it to Market all of the existing video Industries so think like agencies production companies um basically anyone who's like already working with video that felt like the most natural place to take a technology like this right yeah and we like mild success um we did like almost a million dollars in Revenue I think in like 18 months which for like a 10 person team was a significant Lifeline for us yeah but it became very clear to us very quickly that as much as it was very cool it wasn't the path to build a really big company because ultimately because it was like it was a vitamin not a painkiller right the technology was so uh immature that it like took two phds two weeks to create like a 30 second clip right so we did like David Beckham and Snoop Dog and D Messi and some of these big celebrities and those were like you know fantastic uh projects for us to do as a company but they are not like a great business necessarily and it only worked if you look directly at the cam right at all times right there's so many constraints around it and ultimately I think what we figured was like all these people who are already making video they're not that interested in this new kind of tech for them it's like yeah it's cool but if we disappeared the week after it you know it would be like we'll translate videos the way before we started but what we learned in that period was because we just went out and talked to everyone we could about video was that actually there are like Millions maybe even billions of people in the world who are desperate to make video today but they cannot make video they aren't making video today and those are not people who work in cool agencies and production companies they work in like a big Bank like a smaller company they're individual creators and what these people were telling us was that we would love to make video we really because I have this important message that I want to share with my sales team or my executive team whatever and I know that if I delivered as video it'd be so much more effective but I can't get the budget I don't know how to operate a camera and so left with like sending slide number 2,600 uh to these people and it's just not working and so the thesis kind of shifted from let's build tools with the video industry to saying what if we can build uh a way to make video which is a thousand times easier and a thousand times more affordable than running around with a camera which is very unscalable for most people and so that what that is what turned it on to this you know Avatar idea SAS platform super easy to do sales service and that launched in 2020 and that was the inflection point of the company where I think it's I mean I've been involved in like lots of startups over the years and it was like a holy there's something here the product like barely worked the videos were like really crappy the voices were crappy it was like barely working yet we had like a lot of people saying hey this is like doing something for me because you guys have continued to grow a lot how much do you think that is due to the technology has just gotten like a lot better versus the market around you has just matured more and more people have seen the benefit of the technology even with the existing limitations and so they're more like ready to try it and adopt it how do you think 100% both right I think like if you look at the technological progress I mean just the avatars today and the process of creating your own avatar yeah so much faster and better than it was when we started out and the futility of the Avatar is also you know much higher and we're about to launch some new models PR soon is going to be even better I think it on the quality of the avatars Fidelity of them right like given how fast we're growing down how many use case there is for like this level of quality right when I look at what we're doing today I'm like we're like 10% into the road map you spoke a lot about the

matching like quality with use cases you know back then it was okay this specific type of buyer use case the quality we think we can deliver will like meet their needs are you finding now cuz you guys just launched Synthesia 2. 0 right improved avatars you can more easily create your own avatar you said you have uh more model improvements coming how are you figuring out like what the new use cases are that you guys are able to support as the technolog is getting better so I there's two vectors on the technology it's like the pure Avatar quality and then there's like all the platform we build around it one thing that's also very important to me and to the team is that yes the Avatar models are very important that the core of our product for sure but like the platform is so much more than that and the reason our customers love us is because of the Avatar models but it's also because all the additional workflow and edits that we've built around it I think ultimately we're at a scale now right where we can let the users tell us the beautiful thing is like we always express one which is our first emotional avatars and we can just see like people come to us and say like hey actually now I want to do like more product marketing want to do more like external facing content as opposed to some of these more internal facing content I think the really big way of thinking about this is like if you think of where we are now I think we're great for instructional content but not storytelling content yet yeah instructional content to me is like it's very practical it's utilitarian it's about it's not about like you connecting with the avatar or like laughing at a joke or it's just like you know you're teaching someone you're teaching a developer how to make the best use of assembly Ai apis and for the developer that's his or hers preferred method of ingesting that information and that's great but it's not because they're like wow this Avatar is the most funny it's not you know yeah I think what we've learned over the years one of the I think most profound insights that we've had is this thing that most people really don't like to be on camera they don't like watching themselves on camera they like listening to their own voice and that's very through in the a bit more you know like business Centric Enterprise use case where like a lot of our customers is like well people don't want to make screen recordings of themselves even down to that they don't want to be on camera like they don't like to do it and they're not good at it takes a very long time to film because you have to do it 10 different you mess up you have to restart yeah exactly but there's also what we've seen a lot of people who want to create like you know awesome content online but they're not comfortable sharing their face either just because they don't want to because they a they don't want to be like a famous person right but they have a lot of potential to create really cool content right of course you've seen this like somewhat play out with like faceless YouTube channels and Tik Tok and Instagram but I think ultimately what it enables right is that the best ideas can win like if you're a great conent crate doesn't matter how you look uh I have a Danish accent right like uh I don't know if that's good or bad if I wanted to become a Tik Tok influencer but it has some impact on how it would be as an influencer right and so you can take out a lot of those things people who don't in business context don't want to have an accent for example right so it is going to create this like a lot of like artificial content it's going to be weird but I think it's uh ultimately I think that it's going to unlock creativity in a way that we can't really fathom yet what do you think are

the differences in running an AI company versus running a just like traditional SAS or just any traditional type of tech company we could sit here and talk for hours about where the technology might be in 5 years and 10 years the timelines there are unknown the impacts are unknown the reality though today you guys are delivering value for crazy amount of customers there's a ton of use cases where the technology is good enough today and it's getting better that's unlocking more use cases what do you think are the differences between running an AI company uh AI first like AI native company and then traditional SAS company I think it presents a lot of like interesting strategic questions that I feel like we're running an AI company right as opposed to running a SAS company like where in the stack do you play right are you building true foundational models do you take true foundational models that's open source and make them your own through lots of intensive Heavy R d do you just use apis or like wherever on that stack you kind of sit I think you have to be intentional about that but also accept that it can change over time right so when we started the company I there were nothing right there was like literally nothing to go from we had to build everything ourself from like the facial Landmark detections for example which was like a big important part of like the first iteration of the technology now in a world where like there's so much stuff that's happening online we're shifting to all these like generalizable models a lot more people are just thinking about a ibdu EXP with a and so I think a lot of like the hard questions that you face is this thing of like when do you chase the shiny new thing right where like okay clearly this is going to be the future of the space we you just like jump ship to this now when do you kind of hold your guns and I think we've definitely made mistakes in both ways but sometimes we've been too eager and conservative and because there's so much uncertainty with these things right I think that's like a bit different than SAS where I think if you're building a fintech company like if you can imagine it you can probably build it and that's not to talk that down but for something like this right it's like I want to make videos of an avat two avatars sitting in a chair like we are right now and be able to fist bump right like yes that's going to be possible someday but there's a lot of fog of war in terms of like how exactly do we get there right whereas if it's like a fintech thing where like we want to support crypto there's another set of objectives but you can lay them out a lot easier which means you can plan for it better and probably also mean you can execute on it faster right

I think also like what is the primary value driver why is your company valuable um I think you know when we started out like the first couple of years probably most of the value for us was like the models we build right the technology in itself and I think where we are now as a company is that the a models are definitely a big and very important part of like our value as a company but we've built so much stuff outside of it right all the boring stuff like permissioning being ISO 42,1 100 compliant like the publishing platform like all these other things together they are very valuable because we've built the work flows right and I think some companies will I think for a lot of companies the value will be in the models and that's like that's the right thing to do and then you need to make sure you're really winning on the models because that's what you're betting your company on I think for some like us I think probably the value of the models should decrease over time right and that's not because we don't want to do like very deep research but we building a company in a way where the value should ACR from the totality of the platform not just from uh from the models itself so how do you then as the founder CEO think about updating your mental model on where uh in your stack you think is strategic for you to actually invest do it inous rather versus okay let's use an API or let's bring something in open source or partner like how do you just think about what to invest in and then what is strategic for you to to uh do yourself versus just like partner so I mean first of all being surrounded by like an amazing team of people who are much smarter than myself and who actually you know are in the code and in the models themselves is very important like I'm not a technical founder uh just to make that very clear I think the other part of it is really in some sense elevating to like a business strategy question what do we want to be uniquely good at Synthesia and how do we what does that mean Downstream for all the modeling that we need to do right well you could say that means we need to have the world's best voice metal model specifically for videos of human presenters talking to the camera and so with a video model maybe with a video editing model that sits again and the combination of all these models are the thing that we think makes us unique right yeah where does that sit in the space of like what a lot of people are doing and for me when I think of like the video space right now of course what we've seen a lot of the last couple of years is like Sara Runway P all these companies doing like amazing work on but I think most people think of like text to video right and so when I look at a space like ours I look at them and I think you know what they're doing is like they're starting at the point of like complete creativity completely free form type anything you want it'll spit something out which is really cool but the tradeoff is that it's very low degree of controllability M it's not very robust right uh it's very unstable and those properties work really well for someone who's like a creative person right you don't mind doing like 10 prompt to get to like a really cool clip you never would have got to otherwise it's almost like part of the creative process exactly it is like you go back and fall with the computer and it's like it's really awesome the problem is if you're sitting in an office somewhere and your boss asking you to create a video about like how to make sure you're staying safe in the local Warehouse that model does not work right so what we're doing is we're starting at the point of like complete control stability robustness and so on right every time you go into theia it's always the same Avatar no matter what time of the day that you generate a video with it h it's always the same Vice it always generates and you're not going to run into all these sort of issues that you have with these bigger models right now the tradeoff for us is that the technology is less creative it's less free form it's kind of more constrainted more structured in a way but I think what's happening right is that we're building towards more free form more creativity I think a lot of this technology will eventually converge at the base level and it'll be the expression of the product um that matter in terms of what company that you're building right so for us I mean we're building for an audience of people who used to make Powerpoints and now they're making videos right and that's very different from to building to someone who may be like film director a creative director who's always had to rely on like real physical cameras to bring the ideas to life and now they can do it from behind their computer right it's very different incentives it's very different pricing models probably it's very different asks for the customers to the company right and I think that's the thing you I think you have to be pretty kind of um cognizant about like who you're building this for exactly yeah it's like you know you said the PowerPoint to video like that's your core audience right now a

lot of times in the conversations in the AI Market it is so focused on the technology okay where's it going to go in the next year in the next 5 years and I think that is fun convers ation to have but what I don't think people talk about enough is like well who's your core like who are you building this for and um who's your core customer and like who's the product designed for because you can create a lot of differentiation uh once that's clear and a lot of that isn't necessarily always on the technology level I mean you spoke about like the permissions there's crazy amounts of collaboration features in Synthesia um that probably wouldn't exist if you were focusing on uh influencers as your core audience I imagine 100% right and that's why it's so powerful when you anry yourself to your customer um you know you they hopefully will give you a good idea of like what the correct road map is and often times right as you said it's not just the model performance you know it's a lot more than that I think in many ways many customers no matter if using assembly or they're using Synthesia right like a 5% increase in model quality of course everyone will take it but let's say that you know someone comes out with a slightly better transcription model but you've actually built the entire workflows for that company it's integr their you know translation systems and like whatever things that they've been building with you guys it's a beautiful API it gives the right error codes back when things are not working like those are the things that really matter in production right and I think it's also why unless you really want to be like just a pure model company of which I think there will be a general purpose very few there and I think in the llm space it's it feel pretty obvious who those are going to be then you need to build specialization on you know whatever use case it is and that is a lot of the value I think that you can deliver to your customers right now there's a

ton of incumbent Tech products on the market that are trying to pull AI into their product to either become AI native or build AI powered features it seems like a lot of them are maybe just starting with like okay let's just plug some AI features in and learn and see what's device you have for either product leaders or engineering leaders or early startup Founders on how they can try to find where to focus and not just on like the general technology application but like actual uh core customers or use cases so that they can build all that product on top of these models or in addition to these models where there can be a lot of differentiation I think you have to be driven by like a deep curiosity first and foremost right of like really exploring like the problem space I think you have to look for those like 10 next improvements MH has to be like a step change for the end user right in terms of whatever metric that you're messing on in terms of like cost or ease of use or whatever and I think that's exactly kind of our story right like we went all to all these video companies this is kind of like marginally interesting to them yeah but then we found out that actually there's this like whole group of people who you would never have thought of going to talk to about a video products um that had this kind of insight and we heard that enough times we feel like okay there's like actually something there and it's very critical right because had we not been very open-minded and I think you know just really try to build like a first principal to understanding of video we would probably have ended up building a super complex visual effects pipeline tool for making these AI video translations there's a book called The Mom test that um I don't know if you read it but it talks about like how to interview potential users of your product and it's called the mom test because anything you say to your mom she'll say it's a good idea right and so what you want to try and get out of people is like not the mom answer but the actual answer yeah um I think that book was always do want we referenc a lot in like the early days of really trying to build that first principles understanding so you guys are growing a lot making a lot of noise now there's competition that's entering the space how do you think about maintaining your Competitive Edge is it the technology is it how everything's packaged together is it like a focus on core user segments that you think are different from competitors what's your perspective there and really with the framing of like advice for all right people have found that 10x capability they're building their products doing well and then they start to see that competition in the space especially as the technolog is getting more and more democratized so people can launch competitive products potentially more easily than they could a year ago how do you think about that so I mean it's definitely interesting because like it's way easier to copy than invent right and I mean we're seeing that like you know we have a lot of companies copying us like one to one like literally copied our API documentation and like all of our blog posts and like and I think that's like I mean that's a fact of life it's going to happen right that's like capitalism is brutal right any competitive manes just gets competed away in a split second you know I think it's also in some sense we had like the market for ourselves for two years which was great for us in many ways but also there's like nothing to like bounce up against and so that way I think like look at your competitors like see what are they doing well right they're probably doing something well if they're growing fast at least I think there are two sides of this question is one is like on the pure AI side and then there's one other more product on the AI

side I think it's very clear to me that AI moves in these like step changes you want to make sure that you're like progressing along with those and of course if you can be one of the first ones to utilize one of these like new Step changes in the technology right doesn't need to be that you invented but if you have a specific use case like we have that you're the first one to like put that out to Market and one thing I see a lot in the AI world right now is that because AI is such a captivating thing for like the mainstream audience because it's so viser it's usually interactive you can play around with it it's like there are these moments there's like the first company that does X y the first comment does see and doesn't mean that if you build the same thing after that it's like has no value but there are these huge like viral waves you can catch if you're the first ones to do something really crazy but then of course when it comes to the actual product I think it's also very unforgiving and the sense that as soon as you come up with something that's a lot better everything else before it looks really on our product side I think is it's about building for specific use cases right so you figure out that people love to use your product in a sales context okay great let's build a workflow for that or people love using it l context let's build a workflow for that but ultimately I mean an AI business is a business like any other business AI business right now you get crazy multiple you get all the hype because it's not a well understood Market but the better understood a market is I think the more you kind of from a perspective revert back to what a good business look like which is like free cash flow there's a lot of talk about um Tech startups AI within Europe within the EU how do you think the EU can foster more Innovation so that there are more startups taking those moonshots I mean you're from Denmark you live in London a lot of the startup Community right now the EU Community is talking about like competitiveness in the EU how do you think about that as a EU founder I mean I think it is unfortunately still way easier to build a great company in the US I have zero doubts about that it's where the capital is it's where a lot of the talent is and just the ambition the mindset is different than is in Europe right I think especially like Mainland Europe as in like the EU right now I mean being from Denmark right and I'm very worried I think regulation politics talk I mean Europe has all the AI regulations in the world now and none of the AI companies right which is I mean just mind-blowing I think the UK is has a really good opportunity to become a very important uh Power in the global AI ra but I also think that there's some quick fixes for sure like regulation taxes how we employee options and like a lot of like practical things that for sure can change um but I also think that there is this like diffuse cultural mindset thing right where in the US I think people encourage you to think big take a crazy bet um and not just in words but also investors like if you have a convincing compelling vision and idea the risk appetite's there the risk appetite is there right and I think we had such a difficult time raising the first round I'm pretty sure that we've been on the west coast it would have been 10 times easier right we powered through and actually happy that we're still based in Europe I want to build a company that's based in Europe and become like one of the global dominant players but the only round we've raised with the lead investor in Europe is our series C right like every other round up until then has been led by American funds and American investors yeah and that's probably because that the serc you know now we're a business and you can understand our business through the lens of like a spreadsheet and numbers and you know we growth stage but that Al always makes me a bit sad that we've always had to go to the US right and find invest a lot of companies whether they're incumbents are just starting now there's tons of Founders starting new companies right now tons of startups being built is there like one piece of advice you think you'd pass off to all those new product leaders Founders that are just starting right now that you feel like you've really learned over the last 18 months two years especially as it's just gotten like so much crazier I think it's a huge generalization but I think if it's obvious it's too late um in many ways there's of course great business that can be built on just like much better execution um but I think if you really are like a technology company that's what you want to be then I mean I think for example like you know I wouldn't start an llm company tomorrow I think the race is off and you really have to be make to to feel very confident you have some like something that's like fundamentally new right but I think it's just like hype is a great tool it's very powerful in terms of like the go-to market motion and the growth motion um but focus on utility not on novelty is what we usually say in Synthesia and I think that has generally been like a pretty good n star for us to not get too caught up in the hpe and remember that no matter how cool your product is at some point in your journey right you need someone in a procurement Department um to review the business value that you're bringing and say okay that makes sense to pay for um you can get away with a lot in these hype Cycles there will be a lot less scrutiny but at some point like uh you know if you're not actually delivering value you're going to yeah the good hype to capture then cuz I like that utility over hype is you want to capture the hype around first to some amazing new utility that you're bringing to Market when you're trying to do something crazy like really big or very interesting the people who talk to you are very different than if you're building like a payroll company yeah right and we felt that firsthand so we've gone through all these like hype Cycles also before the genni moment where you can actually knock the door down to like Executives and really big companies because they find what you're doing interesting maybe even on a personal level and that's so valuable because you can get in you can really understand how they work how they think um how they think about their businesses and what they want to improve in a way that's difficult if you're doing something that that's maybe less exciting in a way um especially like I mean back then we were like 27 28 years old right so very difficult for us to kick in the door to some of these people and I'm very grateful lot of those people taking the time to chat to us in the early days yeah for sure and then use those insights to then figure out okay great dangle the keys like that's the exciting thing right but then also trying to figure out like okay what what's actually valuable for them right well thanks so much for sitting down and chatting this was awesome uh it's always great to spend time with you and I'm excited to see the launch later this year thanks man appreciate it thanks