This AI Rips Your Face Off

+ Add to

Before the recent AI craze, Dima Shvets and his cofounders invented Reface, the viral tool that lets people rip a face from a photo and replace it with theirs.

He says it’s doing tens of millions in revenue and was downloaded by over 270 million people, but it only happened because of a huge failure.

Dima Shvets

Dima Shvets


Dima Shvets is the cofounder of Reface, the AI video & photo generating company


Full Interview Transcript

[00:00:00] Andrew Warner: hey, they’re freedom fighters. My name is Andrew Warner. I’m the founder of Mixergy, where I interview entrepreneurs about how they built their businesses. And joining me is someone whose app you have probably played with. Even if it’s not directly, then it’s probably because your friends have sent you videos using his software or because there’s a collection of avatars that were created using his software.

Schvitz is the co-founder of Reface. It’s the AI video and photo generating company. You can take your photo and they’ll give you these beautiful avatars, or they will turn turn it into a music video, or they’ll change it in one way or the other. That’s the idea here. It’s done phenomenally well.

I invited ’em here to talk about how they did it and we could do it. Thanks to my sponsor, Later on, I’ll tell you why. If you’re hiring developers, you gotta go check ’em out. Dima, gimme a sense of numbers. How many people are using reface.

[00:00:48] Dima Shvets: Andrew, thank you for having me. For now, reface is being used by more than 270 million users worldwide. Since we launched the app in early 2020,

[00:01:03] Andrew Warner: And then how much revenue comes from that. I noticed when I used it, it’s basically not useful until you pay. So what kind of revenue?

[00:01:11] Dima Shvets: It’s free, so I couldn’t disclose revenue numbers. So it’s like a couple, like of something. But generally the app is free. Depending on the product, because we face is not only app we have mostly you can use the most of the functionality of the product. But yeah, the core features and the core functionality and a lot of tools are available under subscription.

[00:01:38] Andrew Warner: So can you gimme a ballpark about revenue? What are we talking about? Tens of millions. Single millions. I’m not looking for an exact

[00:01:44] Dima Shvets: I guess tens of millions.

[00:01:45] Andrew Warner: Tens of millions of dollars in revenue. That’s phenomenal. You know what, I wonder if maybe I fell for one of those splash screens that I tried to create a second or third video, and then I saw a thing that said 3 99 per week, or I forget how much per year, and I said, all right, I’ll do 3 99, and I signed up.

And so I just assumed that I needed to pay, which is maybe one of the flows on your app.

[00:02:06] Dima Shvets: Yes. Yeah. So there is specific reasons. Usually we have a pretty sticky audience, a core audience, and they are mostly paying us subscription. And most of the technology features we have are really cost heavy in terms of, video to video and the latest feature you have seen in Reface and Restyle.

So definitely we need to be sufficient in terms of what we are

[00:02:30] Andrew Warner: the expense in this? Who, where’s the expense in regenerating these videos?

[00:02:36] Dima Shvets: Different. So the technology underlined which is basically was face web. It’s pretty tough to do, and when we started it was really scalable, conditional, and you need to always process all the videos through the service. So it definitely needs a computation on the side, going to the, for example, Google Cloud, and you need a lot of stuff going on to support it.

[00:02:59] Andrew Warner: But it’s all, it’s your software. You are not, it’s your software and you’re just paying for heavy compute power in order to create these videos.

[00:03:09] Dima Shvets: Yeah, correct.

[00:03:10] Andrew Warner: Ah, and so that’s why, for example, the avatar is gonna take 45 minutes for me to get my avatar created once I upload six of my photos. It’s because it takes that long for the computer to process and you have a big backlog of people what’s going on behind the scenes.

[00:03:25] Dima Shvets: Yeah. You’re correct, Andrew, for now. So I believe that our computation power and how we wrapped up technology is one of the most efficient and problem, but in the meantime, it took plenty of time to process it, to prepare and to get it to the user. And we have the solid backlog of users that want it every day.

[00:03:44] Andrew Warner: So what’s the long term vision here with this? Is it that yeah, what is the long term vision? Because at some point it’s cute to put my face on someone’s music video, but after a little bit the cuteness wears off. There’s gotta be a substantive mission that you started this business and built it up for, what is that?

[00:04:03] Dima Shvets: When we started the company in 2019 our vision always was to empower content creation using technologies. And that’s the vision and the mission, I believe, like the mission that always help us during our tough times. Yeah. And we want to empower everyone to create amazing content. That was the intention why I joined, started the company.

My Why. Micro founders did a lot of technologies before a phase associated with

[00:04:29] Andrew Warner: by content creation? You mean these fun type videos? It seems like you have a bigger vision in mind, more than just having Abraham Lincoln sing a song. It’s something else. What is that? Yeah, what is

[00:04:39] Dima Shvets: Our our vision not only regarding content, but general, is to build an outstanding generative AI company. Because what do you see now, Andrew, in the world of generative ai, text to video, text to image, text to text, video to video, reface, and our micro co-founders and current co CEOs and vision even before when we started the company three years ago.

So our base vision was okay. If we can create the technology when everybody could literally be anyone, but changing not only faces, but not only making them speak, but not only changing their ground to make different technologies work on based on your voice. Our intention was guys, there is a huge avenue of people.

Who are limited of creating content at that time because of the technology, because of the costs, because of your computation, because it’s good. It was really tough to get the technology work in the way you want. You need to go to post-production studios at that time. You need to work with specific, content creation studios to create the ads.

For now using Reface and a lot of prominent companies that are now booming in generating, you can do it on the fly. So what we started in 2019, as I envision that though, you don’t need to be Instagram star, and to make your offensive photos doing some stuff, you can just take any piece of content and recreate it.

Like monolith. Yeah. If for example, you go

[00:06:03] Andrew Warner: the Mona Lisa part feels like a cute, I’ve seen it. It’s amazing the first time and then after that you move on past it, it feels to me like what you’re saying is. If Andrew is a good cook but does not want to be on camera because he doesn’t have the camera work, the editing chops or even maybe doesn’t look pretty on camera, he shouldn’t be held back by that.

What he should be able to do is just go to one of re face’s, apps, create it using somebody else’s video that he manipulates using his own voice, et cetera, and then he’s got his video. Or if Andrew wants to teach something based on an interview and doesn’t want to go on camera, He gets a prettier, better looking person in a shot that he’s looking for.

That’s the vision long term. But instead of starting with that, you said, let’s start with something fun and viral. Am I understanding the mission?

[00:06:49] Dima Shvets: Yeah, correct. Yeah. Yes. Thank you for unpacking it, because it was really tough at the beginning to make people use the technology in such way that use now. Now people understand that you can recreate the content, you can change the voice, you can make judge PT stuff. But at that point of time, we need to create something so easy to use.

Eh, natural for a person that in two clicks you can see the power of ai. That was our, evangelist mission at the point of time.

[00:07:18] Andrew Warner: I also get the sense that, truthfully, today, this technology is not where anyone wants it to be. No one’s watching this and going, Hey I am tricked by it. I feel like I’m watching Andrew really dance. No, there’s more like, how did Andrew get to dance on top of this woman’s body? And then there’s a right, and so you’re leaning into the fact that the technology isn’t ready yet, the audience isn’t ready.

So let’s make it fun today and grow with the technology and with the audience’s expectations. Okay, I, I see now where this is going, considering that you’re on a right path, finally, the world is catching up to you. The fact that you even have AI in your domain name means that people are in love.

Why’d you walk away from the business?

[00:08:03] Dima Shvets: Oh yeah I guess it’s the most, questionable story and the main question I get all the time. For me it’s I talked a lot, so I definitely recommend so a lot of people who actually writing a lot about that, I will tell my personal story. For example, mochi, who is speaking a lot about mentorship and about the shift in the position.

So my intention was that I’m Ukrainian. We did a story which was super and still super viral. In a time where technology wasn’t so popular and nobody even believed that it could happen. So scale and viral, we did our job. We gone through a lot of tough times and and sudden from going super viral, raising from a 60 and that getting not solid retention.

It rate on different products, technologies going ups and downs, different Strategy shifts, pivots, then the war started. Yeah, we didn’t fire anybody from the beginning of the war because of the shortages on layoffs, we optimized the business for profitability and multi-product strategy. And in some moment of time I realized that for me personally, I want to revision how I live my life.

I think I did a great job in terms of building a company of 200,000 people and solid results, and now it’s solid operational, multi-product strategy company. But in the meantime, I have a wife, I have a five year old kid. I have a lot of vision regarding, so what industries can be disrupted by AI in a good way?

Because even studying phase, we did a lot of. Factually navigating people through the power of ai, but still thinking about regulation. Let’s, for example, what we do is not deep fake. Yeah. It’s phase fault. What we should do to make it properly, to make it for good use. What is, what’s about how we can guarantee that the technology somehow, from our perspective, will be done only for good way, not in a bad way.

So all these questions I was. Asking myself during my life in terms of what’s the next evolution of me as a person, as a husband, as a father, as a Ukrainian. That is why now I realize, okay, for now, reface in a good, stable position, still continuing and pursuing our goal of empowering content creation. But in the meantime, I wanted to change my focus to family, to risk, regroup, restructure my thoughts, and start thinking of other AI related stories and endeavors that matter to me a lot, especially.

[00:10:44] Andrew Warner: Yeah.

[00:10:44] Dima Shvets: Yeah, especially in terms of what I envision to be really important in the current and future world, like education, the future of education, the future of health, mental health, especially defense tech as well, meal tech as Ukrainian. So I believe that a lot of problems could be avoided. Not, I don’t, I’m thinking about War of Ukraine.

Geopolitical shift. We’re still in the process of it. Yeah. Global geopolitical shift. But I believe that if we work a lot on the regulation, leveraging the ai, leveraging technologies, not being dependent, from different natural resources stuff like it would be much easier and much proper to develop our world.

And I think that education, healthcare, mental health and and defense, tech for now is the industries that should be one of their. Most focused on in the current world.

[00:11:39] Andrew Warner: you’re gonna pick one of those and focus on them.

[00:11:42] Dima Shvets: Yeah. I’m already advising, working a lot making my researches, advising startups in that field. And yeah,

[00:11:52] Andrew Warner: With the idea that you’ll eventually pick one area and then go focus on that.

[00:11:57] Dima Shvets: Correct, correct one area, but in the meantime leveraging my experience in AI to help other areas to develop. Yeah, because I believe one of my narratives in my life, what I believe is you can be like a midfielder on the soccer field, so you can have some, you can connect the right dots between investors and entrepreneur, entrepreneurs, experts to make big sense that matters.

That’s the story of what I envision, what I believe in, and what I try to execute.

[00:12:27] Andrew Warner: Okay. All right. I’m seeing what your vision is. Let’s talk a little bit, my goal here for this interview was to find out how you built a briefcase, but let’s talk a little bit about the vision for the future, cuz I’m fascinated by this. When you think about education, you’re touching on a topic that I’ve spent some, a lot of time, years of my life focusing on here.

It’s been entrepreneurial education, but I’m curious beyond that, how do you see AI being useful today? And then what do you think five years from now in education, what’s the vision?

[00:12:54] Dima Shvets: My vision is so we have different systems of education in different countries.

And education is definitely affected by religion, the government, and a lot of obstacles and circumstances that exist in specific country or even ethnicity and geography and my beliefs that the future of education should be done on the basis.

Let’s call it, of not national status. So I believe if we could create a system where people will be, let’s call it like , but we’re thinking about the future. Yeah. Of

Where the ST system will be built not on the base of nationality or religion, or on the intersection of different participants that are representing the specific group.

Of ethnicity, religion, nationality in a specific way where some decentralized body that controls how education being done. I believe that we can spread the education across the globe. Not

[00:13:52] Andrew Warner: But that’s like something that the internet could have done and has done. There’s something about artificial intelligence that you think is going to change what we couldn’t have done before could now be possible. What do you think? How does AI play into that?

[00:14:05] Dima Shvets: I think that AI can significantly accelerate how it can evolve, for example. Yeah. There is specific it’s called Two Sigma Problem. Yeah. Where you can see that individual education is much more reliable and efficient in the way how children.

Children’s students are being educated and tutored, and that is why we have a year like personal AI teacher, personal AI stein, or personal ai, in Newton can definitely help you to make it much faster. What I mean, I believe that natural studies like physics, A sub studies of physics as like mathematics and as they like chemistry are super important for the understanding of the fundamentals of the world. And I see now that a lot of knowledge, scientific bracelets are not coming to life because scientists are not being short, kids are not being tutored, are not being educated the way that they can understand the world, understand the basic, and then can grow and spread this world.

[00:15:12] Andrew Warner: All right. Let me pause here. I think I wanna make sure that I’m understanding you cuz I’ve had this thought too, I’ve said for years now that if my son is fascinated by say, by say gardening, I. His math should relate back to gardening. If he’s fascinated by business, his math problem should be related to business.

What if Johnny has 10 lemonade, 10 cups of lemonade, and he sells five? How many does he have left? What if he multiplies that by 30 days of the week? How much if he makes, right? If that’s the modality that he has, let’s increase our let’s give him topics that relate to that. It’s hard for a teacher to do that, but AI could customize it.

No problem. I’ll give you even a better example. I learned languages, not using repetition, but using mnemonics. And for years I’ve been creating my own mnemonics for each for each word that I’m trying to learn a different language. When I went to Brazil last week, I said to Chachi, pt, you’re my tutor.

Give me 10 phrases that I need to know here in Brazil. And then once I liked it, I said, now gimme mnemonics for each one. And then I adjusted the ones that I couldn’t, that I couldn’t remember, and then we got into a rhythm of mnemonics for each thing.

[00:16:25] Dima Shvets: I love

[00:16:26] Andrew Warner: like that, right?

[00:16:28] Dima Shvets: Correct. Yeah. Thank you for structuring it in a more, practical way and agree. Yeah. So I’m speaking about that as at least at start. Yeah.

[00:16:37] Andrew Warner: Okay, then take it from there. So that’s that to me is not a today solution. The today solution is someone using chat, G P T or a similar system to create something where you’re basically teaching the teacher what to teach you. What you’re talking about is a year from now, maybe two years from now, no doubt we’ll be able to walk in into a chat room that says Andrew needs mnemonics.

Andrew needs to know this next thing on the syllabus, we’re gonna combine his mnemonic needs with the syllabus that someone at his state needs. Talk to me about a little bit further in the future for education. What’s possible?

[00:17:10] Dima Shvets: And yeah. I agree with you and thank you for this example, Andrew. And one of the angles that are interesting for me as a assumption, Even going earlier, for example we are speaking about already school education. Let’s teach math. Yeah. In this specific way where a kid or person is eager to develop a him or herself.

But what about early education? Kids who are fans, like of different characters, movies.

Cartoons. What if in, the power of technology you can create specific characters that can engage people, can engage kids? So my perception that I have a daughter and what I realized that you need to help your daughter to love something and to be curious, continue to expand the curiosity along the life.

And if I understand that she likes El. From Frozen, it was easier for me to get her know numbers when Elza is speaking to her. That’s the evolution that the specific support with AI that can go. Alongside the way, not only from the school, but starting from three, four years till the university sometimes, and we should leverage the technology and, but next what?

I believe that if we make people, and we, if we give people opportunity to get this fundamental knowledge in a more engagement, playful way, that means that we’ll have more scientists. Because I believe that world likes real scientists. And there are a lot of guys whose ideas and breakthrough technologies are not coming to life because of different, business environment interpersonal stuff and communication, which is usually, the treat of a scientist.

And I have a lot of friends, Ukrainians, not only Ukrainians in different universities and one of the company in like in defense tech, which I really admire. They call it like necessity is a maso innovation. And those guys created the audio detection, narrow metric because of the necessity in Ukraine.

And they have a close community of scientists that join the forces and they call it asymmetric technologists. And if those guys wouldn’t gather together because of necessity and the unique situation in the world, those technologies might not even seem their life. Might not even be on board.

So my intention, if we do it properly in terms of education from early days, and we create a system where Stanfords of the future will be developed across the globe, that means that we’ll have better future for our kids. Yeah. That is a vision. Like how we can do

[00:19:50] Andrew Warner: So then what do you imagine for scientific education?

[00:19:55] Dima Shvets: I think that it’s already happening, but we should create like network effects for scientists for my perception as a professional in business or like in sales half we have linked in. Yeah. And those guys, you understand. Second, third tier connections here can interact, experience, share your knowledge, gather the feedback.

We should think how we can get. Those guys together in a way that is comfortable for them, where they can make breakthrough technologies and breakthrough scientifics come through faster.

[00:20:27] Andrew Warner: So the early part of the internet allowed scientists and academics to communicate with each other, and by communicating with each other, they helped each other. Learn more, teach better. And really they became smarter. Then more data goes online. They were able to, the people who came on the internet were able to tap into that and all the way over to even gamification that allowed people to learn better because we were starting to learn game mechanics and what drove people, and we used it to teach them all.

That is good. The challenge with AI is where, with all of those techniques that happen, up until now, there was a lot more guardrails. With ai, it’s really hard to predict how it’s going to teach, especially if it’s customizing infinitely. And so I think about kids. I don’t think I would feel comfortable with a kid learning from Elsa that was AI generated if I couldn’t see what Elsa was teaching and make sure that it’s accurate.

And how do you deal with the lack of accuracy? Especially with kids, and especially when technology requires a lot of experimentation.

[00:21:34] Dima Shvets: Definitely great and reasonable assumption, Andrea. So I don’t have an answer for now. What I can suggest and what we need to do. So we need to simultaneously work on the regulation and simultaneously work on the frameworks and the technology development. Like even Sam Altman’s speech INS eight, I think those Sam Altman’s of the world should.

Speak and work in Europe, in Asia in Pacifics across the globe to think what is relatively good and what is not. Because at that particular point we couldn’t foresee and forecast how it’ll develop. We’ll definitely see what can be done good and what can be done better, and the matter of the reasonable trade off.

Socially informationally, even commercially and regulational, how it can be developed. That is why that is the specific issue. That is a challenge, but I see more, much more, economic, social, not an economic, social and cultural upside it than downside.

[00:22:41] Andrew Warner: I still don’t see how the path to experimentation can work and get us there. And what I’m imagining, that I feel like you and I are different types of entrepreneurs, and I wanna learn from your approach because my approach is let’s put out a minimum viable product. Let’s have people experiment.

Let’s keep learning and innovating, and let’s keep testing. And with something like this, with education, I don’t think people are willing to experiment to that degree, to the degree that even like I’ve been trying some ai podcast editing content editing tools. They’re fine. And when they make mistakes, nobody dies.

Nobody ends up really in trouble when it comes to education, military and and science, as you’ve talked about, there isn’t that M V P approach, so how does DMA approach it?

[00:23:26] Dima Shvets: In my perception I think it should be combined. So definitely like how we build companies and how they’ve been built. It’s like MVP approach for sure. But in the meantime, I think there is some challenges, especially for example, in defense where I started from that in could, couldn’t do just mvp.

You should have a broader vision, what the world would, might be like. You should have allies, you should have specific frameworks which go beyond the mvp, for example, in Ukraine. Ukraine is a unique battlefield for now. Yeah, and the feedback loop on the innovation and the software is going super fast. But if you don’t have exact strategy, how the world might look like in five, 10 years.

That will be extremely painful, possibly painful and extremely stressful, for example, for Ukraine to live in this current situation. So my approach is I’m definitely not a fan, but that’s important part is MVP and prototype and how it works. But I’m more, trying to think visionary strategically what was done before.

What is not done and what is the intersection where we can, for example, education. Yeah. We see the system that’s been developed. I know the system of education like in Ukraine and a specific system of education in uk, but I try to understand, okay, beyond the regulations, beyond the where we are technologically, what might be the unite force for that education?

What might be the amplifiers of education for kids in UK, in Ukraine, in and urban world. That is why I’m more like kind of approach, which is more like might be considered vague. Yeah. And visionary, not so practical, but antra always to go zoom out, zoom in. And when I see that something triggers my mind and my dataset and people who aligned with my vision, scientific businessman, regulators, I try to dig deeper and double down on this situation.

[00:25:26] Andrew Warner: I see. So it’s not like you have a systematic approach that says, first we launch a mi, a minimum viable product, then we put it in the market, then we improve. It’s more I’m gonna keep my vision out there. I’m going to keep talking about where we need to be and what the ideal, perfect world is not mess around with the in between where we are now and there, and then I’ll see who gravitate towards me.

Who’s done this or who can build it or who has better ideas. And then you’re just gonna keep pursuing what works and pursuing what’s exciting and pursuing what gets you closer to a vision that matters. And the mechanics of how you do it are not as important as, is the vision clear? Are you drawing the right people in?

I see you’re trying to interrupt in between, which tells me that I’m getting it a little bit off. So correct me where I’m off.

[00:26:10] Dima Shvets: Yeah, like you got the main point for me now. Maybe it’s a state of my mind for now. The state of my, situation. I try to focus on who and why, and then only on what. So I did specific businesses before I did VC before Rephase and Rephase. So it’s all about iterating. Yeah, iterating and testing.

A lot of sis, user testing, different apps. But after that, I now change my perception and my strategy of life to the point like why, who, and then what.

[00:26:44] Andrew Warner: All right, so let’s see how you did this before. In fact before we get into Reface, I want to tell you my sponsor is Lemon io. They’re actually a they’re a Ukrainian company also, and for a long time, Dima, I was saying to people that this Ukrainian founder built up this phenomenal business where he was matching business people who are looking for developers with developers from Ukraine.

It’s If you’re looking, I might as well get credit

[00:27:09] Dima Shvets: Yeah. Yeah.

[00:27:09] Andrew Warner: over. I see that you’re going over. Phenomenal entrepreneur. He kept talking about how his goal was to hit a million dollars and he talked publicly about how he was getting it and then 5 million so on. And then the war happened and the whole thing got derailed and he said, okay, I’m gonna keep paying for my people while I figure out how we can build a business.

We can’t get developers from Ukraine. So we started looking for other places where they’re phenomenal developers who are not as highly paid as they are in Western countries like the us. Anyway, I talked about that. And then Dima, what The reason I’ve stopped is a lot of the feedback I had to that ad was related to Ukraine and not to the fact that he has this phenomenal service and that he’s got a good team of people and he can match people with developers who are in artificial intelligence or blockchain or anything.

And so I’ve stepped off of that. So what I’m gonna do is I’ll come back to you and ask you about Ukraine in a bit, because I can see you brought it up a bit cuz it matters to you. What I’ll say to my audience is, what matters to you is you want great developers at a great price by a company that will match you up like a perfect matchmaker and ensure that you have success.

Go to not just lemon io, go to lemon io slash mixergy because if you do, you’re going to get a bigger discount than other people do. That’s what and frankly, I’ll get credit for it. It’s phenomenal because the guy, the fact that he’s so freaking resilient in the face of war, in the face of all these problems that he can keep his business going and growing means that he’s gonna find, he’s gonna find my audience a great match.

Go really get developers from g. Alright, Dima, I wanna see how you work in with a practical example. Reface, what were you doing just before Reface?

[00:28:47] Dima Shvets: I was doing venture capital. I was doing investment in early stage startups. Starting from 2014. Mostly in Ukraine. Mostly sit stage. Yeah, that’s how I was approaching tech ecosystem from now.

[00:29:02] Andrew Warner: Through where? I’m looking through your LinkedIn. I don’t see that. Which of these programs were you with?

[00:29:09] Dima Shvets: It was a fund I G V C. So if you can see my

[00:29:12] Andrew Warner: What was the name of the fund?

[00:29:14] Dima Shvets: I G V C.

[00:29:15] Andrew Warner: I gvc. Okay. All right. So it was through I gvc. And then what made you say I need to go out and start a business? I.

[00:29:22] Dima Shvets: It was always my vision, I’d call it that. I always had a feeling that I need to join the bright side of the moon, how it’s called it. So I spoke to a lot of entrepreneurs. The market was still emerging and evolving. It was still premature. Compare it to, yes, to the Western world, but there are a lot of tech, great minds engineers in Ukraine that started building products of outsource.

Yeah. We have a solid outsource out stuff industry out from Ukraine and a lot of people started earning a lot of money, making exit and starting to building products and I think this boom was starting from 20 14 15 when I. Actually started to doing venture investment at that point of time. I always realized that, okay, I’m speaking to entrepreneurs.

I have a management consultant, investment banking background, but it’s not single enough. I’m, to be honest, missing operational experience building tag business.

[00:30:15] Andrew Warner: Okay.

[00:30:16] Dima Shvets: So I was always thinking, okay, what is the story I would like to join full-time? And somehow to sacrifice the life of venture capital guy. And I realized that I’m not going to do some, marketplace eCommerce with all the respect because it’s super tough then to scale it from Ukraine to the western world.

To us, for example, as a main market. I was always obsessed by deep tech and I was okay if I’m gonna, if I’m gonna join, Startup or build a startup, it’ll definitely be a deep tech company because we, I will definitely have some, competitive edge, let’s call it like competitive molds, even compared to, yes.

So we’ll have at least like six months or so advantage to build at that point of time. I was looking for technologies. That’s a tough to build and replicate mostly narrow networks and ai, artificial intelligence. For now, everybody, it’s like a boss. I’ve been following AI for five months as Gen Z, VC Starter Pack.

But first my touch with AI was when I met my current co-founder, always the CT O when they were building, not even Rephase, they were doing 2D to 3D conversion of films. So you know that films are usually even now done, like you need 600 people to make some Marvel movie to be converted from duty to 3d, semi manually. And they came with an engine that gives an opportunity to do it. And I started speaking with them. I was really inspired by what guys were developing and it was like, I guess 2018, even 2017. So I started to investigate those guys using Ganz generative adversarial networks. It was the newest technology of training nano networks, which appeared in 2014, and I started to dig in deeper in the space and then guys started speaking, okay, we’re thinking about the new technology.

Yeah, and that’s how Reface was born because guys were doing the tech, which gave an opportunity to face swap in, in pictures. The technology was not so prominent even close to what we had. But there was one interesting point when they launched the app on for Elon Musk. He used it and he faced what himself with Dwayne Rock Johnson.

And that was, the moment of not glory, but the probability and the moment of understanding that we are doing something right.

[00:32:45] Andrew Warner: When he used Reface. So by then Reface was already up. Take me back again to the beginning. You are saying you didn’t have a theory of where the market was or what product you needed. You just said, I wanna see. What technology is exciting and this team was doing exciting technology by taking standard 2D movies, making them 3d.

And you said, I don’t know where this is going, but if they could pull this off, I know there’s gonna be a market for it. And they’re smart enough that I want to watch them and participate in their growth. That’s it.

[00:33:16] Dima Shvets: Correct. Yeah. And

[00:33:17] Andrew Warner: even, I see a problem in the market, I need to solve it. It’s, I see smart people who are working on smart things and I think they could do it.

I want to be a part of that.

[00:33:26] Dima Shvets: Correct. And what is super interesting, Andrew, you totally got the point that 2D to 3D conversion didn’t go anywhere, but that was the beginning. Why then Reface appeared and

[00:33:38] Andrew Warner: Okay. But still, let me stick with this moment in your time, be in your life because what were you offering them? So they had a team of people who could build this technology. You weren’t gonna sit there and code with them. What was the part that you were gonna play?

[00:33:51] Dima Shvets: I was a guy who were mostly focus on investments and understanding the world of investors and the entrepreneurs from different standpoints. And second the main hypothesis for technology was doing it B2B and selling api. So our main assumption at that point and the guy’s assumption or edema, could you please join us and help us to build B2B and to build the product on B2B and to build the, and to build the funnel.

And I was my approach was okay. You’re totally right Andrew. There was no problem. Specific problem at that point. Yeah, we had this vision. Yeah, it was a good izer, but at that point there was no specific market on that. But we had the technology that was scalable and conditional, so at, even at that point, we could process millions of Facebooks per hour.

So I started to iterating. I started speaking to streaming services, gaming companies, marketing agencies, different kind of businesses, and to start thinking how we can make technology being a product on API basis and on subscription basis, and to mostly get to recurring revenue potential.

[00:35:05] Andrew Warner: So you were talking to them and it didn’t work out. Was it because the customers weren’t ready to buy it or because the technology wasn’t up to the level that they needed at these streaming services and other potential customers?

[00:35:18] Dima Shvets: I think not first, not second, but second suggestion. So we did some contracts on a recurrent basis. This, those contracts helped us a lot later when we got consumer on the partnership side. We did partnership with gaming companies, with streaming companies, with different marketing agencies, and a lot of these connections with such and different marketing agencies helped us them to make a lot of prominent partnerships.

But at that point of time, face swap was viral, but not recurring use case. So you used it, it has definitely blown your mind. It was viral, but after that it was tough to get people to use it. Professional and recurrent basis.

[00:36:01] Andrew Warner: But then how do you get from, we are going to change movies from flat 2D movie experiences to full 3D experiences. How do you go from that to Face Swap as an app that. Elon Musk can use. Why even go to that? There seems like there was something that the market or your development was telling you wasn’t right, and you needed to pivot.

[00:36:23] Dima Shvets: Once again, 2D to 3D wasn’t the basis of three phase. It was the technology, how I got connected to guys.

[00:36:30] Andrew Warner: Yes.

[00:36:30] Dima Shvets: it wasn’t, yeah. And the guys closed this technology because they didn’t find use case or, and on a significant technology level to get this product of the ground. So you’re correct to,

[00:36:43] Andrew Warner: it seems like it’s, if the technology was a lot better, the market would be open to it. But they weren’t. They don’t want something that’s experimental. It has to be so amazing that no one in the audience could tell the difference. Got it. And so then, When they had this technology, how did they, and I guess you at the time think through where else could you apply what you already have.

What was the process of going from this tech isn’t working for streaming services? Where else could we find a customer base for it?

[00:37:14] Dima Shvets: For

[00:37:15] Andrew Warner: Yeah. Like where else could you find, what was the pivot like? How did you come up with this new approach that ended up working?

[00:37:23] Dima Shvets: When I joined the company, it was only Facebook technology. So it was Facebook technology that was scalable, conditional, but didn’t have any clients. And my bet

[00:37:33] Andrew Warner: you are gonna go do b2b. B2B was not working.

[00:37:37] Dima Shvets: Yeah. It wasn’t working before I came. Okay. I believe that. There is a high probability that it doesn’t work because the market isn’t there. The technology is not sophisticated, but I believe if it works, it might be massive.

[00:37:50] Andrew Warner: Got

[00:37:50] Dima Shvets: So I got this binary approach sometimes as usually and went exploring.

So I started getting feedback from a lot of France and streaming kevin Lee, like gaming companies, even in China, like marketing agencies and guys, you may have an interface. You can use it for specific marketing promotions. We give you the engine. We are not making it for you on the, clients and some service basis. We have a technology. You can create all the campaigns for your clients by your own. So let’s do it. But at that point of time, I believe that. The timing, which is crucial go to market on the B2B site, wasn’t there? So the technology was already great, but it was only Facebook and my perception that we already were going to the point where we can make it recurring. What I mean that a lot of people, I call it, digital melia. So my intention, what is going now with companies who create. Creative campaigns for the businesses using ai. The same with us, but we had on the face swap at that point of time, and we thought, okay, all the social media managers, all the marketing creative campaigns are done by just a lot of work of manual work of marketers.

What if you can be like digital memory? So someone who creates memes with a helper face, swap ai background and use it as one of the. Metrics how perform our market and our marketing has been measured. And we already started to kick it off, but to be honest, at that point of time, it was, yeah, it was really nice, interesting business.

But it didn’t work on the scale as it suggested. But another point of

[00:39:39] Andrew Warner: technology wasn’t there.

[00:39:42] Dima Shvets: Because the technology and the use case of the technology wasn’t so repeat, and so it was nice, it was novel, but still not enough that, like the, like instruments? Yeah the instruments that you use it for combining, like you see now possible, for example, in runaway ml. Wasn’t technological at that point. So we did Facebook unique first of the kind, but it was not enough for the instrument that needed for the businesses to create campaigns. For example,

[00:40:10] Andrew Warner: Okay. Got it. So then you start doing this for consumers. Consumers are more forgiving, they’re also much more sharing. The first consumer app you did did how, what was the reaction?

[00:40:22] Dima Shvets: The first consumer app was called Duplicate, and we thought it was a great name in for Yes market. But not, it was double. Okay. It’s double you. Twin, double. No, and we able to rephrase what the company name is and by the, but there was a crucial moment in our history when we launched Reface videos in guess the aha moment started in June. It was the state of mind when we finalized the latest video technology. And second, we created the short form of content, which were called promos. And in our business, like in consumer internet formal content is crucial for virality and organic growth because so many consumer companies then experiment with the form of content.

And what we created with Swiss promos and different ones filmed by ourselves, not only ourselves, was crucial with the combination of technology to go viral. We grew from 10 K D a U daily active users to about two and a half million, like in two and a half months.

[00:41:32] Andrew Warner: What is this? Sh these promos I’ve been using the app. I don’t know if I understand which of the different content types is a promo.

[00:41:42] Dima Shvets: I think what you see all the form of content that you see in your face currently, it’s a promo alongside avatars because avatars appeared like not long

[00:41:50] Andrew Warner: So the promos are these little short form videos where maybe there’s a woman running walking down the runway up to 10 seconds. I get to take her face off and replace it with mine. And then maybe I, once I have that, I wanna share it with someone by iMessage or other social media tools.

[00:42:05] Dima Shvets: Correct. That was the first use case either to share. So show me there all through your messengers and direct communication for the beloved one, for example.

[00:42:12] Andrew Warner: Okay. And then the next one, what was the next big use case?

[00:42:16] Dima Shvets: The next big use case was then you can not only face swap your image that we, that you can reenact this image. For example, you cannot only take the existing video or film video face swap. You can take the photos and to make it speak. Then you can not only make it speak. You can make it speak what you want, not what only what you want, but with the face you want. And now this is like specific question, but what if ever since this world will be like Black Mirror? But no, that is the stuff what we did a lot with different regulations, NSF W detectors, it’s like detectors we have in mind and a lot of policies. And we did tremendous job to be honest, like for almost three years to. Appreciate and to show the audience, the regulators, the businesses across that we face is about content creation, entertainment way, not in another way that can not defect. That was one of the

[00:43:13] Andrew Warner: know that it’s not deep fake is that you’ve got a watermark in it so that it can be detected. It’s like a printer where if I try to copy something that I’m not supposed to like money, they could trace it back to the printer that I used. It’s that kind of a thing that you put in there.

[00:43:28] Dima Shvets: Correct. Yeah. So we had the most obvious part of identification watermark, which you can see everywhere. Another point we developed a lot of tech, which is not being released, but still in places like Invisible Digital Watermark is footprint.

[00:43:43] Andrew Warner: Huh.

[00:43:44] Dima Shvets: That can be identified and for example, if person do some bad stuff, it can be identified.

And when you know that you might gonna be caught and even if the system somehow goes you in anyway, there is some, probability that you might be somehow tracked for that point.

[00:44:01] Andrew Warner: Okay. And then the virality comes from, I guess there’s a watermark if somebody uses the free version. And when people use the free version and post it publicly, that’s how you get your following. Am I understanding that right?

[00:44:14] Dima Shvets: Yeah. Yeah. It is a basic mechanic, so you see is a watermark, and that’s how it’s gone. So it’s gone super organic and viral in the first days.

[00:44:23] Andrew Warner: All right, so tell me what happened during the war for, I could see that the Ukrainian situation is really important to you. I think a lot of people would’ve said, I’ve got it. It’s painful, it stinks. I’m gonna complain about it and move on. For you, it’s even more painful and you’re not moving on. What happened when the war started?

[00:44:44] Dima Shvets: So when the war started I think rephase, the company was prepared for that. So even before the war, we had a lot of US investors and most of our audience and user base and the business in US and UK and Germany, France. That is why we had a lot of, let’s call it business contingency plans and in place.

So it didn’t affect our business necessarily, but it definitely affected the people. Them, their emotions and their relatives, but


[00:45:18] Andrew Warner: kind of what kind of backups did you have? Did you have employees outside of the country? Did you have a way of getting your team out? You did both of those. So if everyone in the country had to go to war, you had other people outside of the country who were prepared to take over wouldn’t have been the same, but it would’ve continued.

Got it. But at the same time, you have employees whose lives are being impacted by this, whose families are being impacted. Sorry, take it from there.

[00:45:38] Dima Shvets: Yeah. And in the meantime, so like we helped a lot. We even launched a face fund, which was humanitarian fund. We supported different funds, which are global. We have Ashton Kucher has investor who helped a lot and gathered some funds with Mila, different guys, different intentions. We sent about 15 million push notifications to Russian audience in first two weeks. To tell them the truth, to give them access to trustworthy media. But it didn’t help. It didn’t help to be honest. We closed operations and we closed this opportunity in Russia, and along alongside one of our cofounders came with proposal upfront to rent some hotels full autonomy in the western parts of Ukraine in case of Black House.

So even six months is before the October’s blackout. We already rented hotels with Starlings and different systems Western in Ukraine to get people out in case it, it goes bad, and it definitely helped a lot. Yeah, and I mean like a lot of initiatives, a lot of revisions we call reface, people re revisions.

Even now I’m step down from my position, but I’m still Reas and we have a unique dna, those guys make an consumer entertainment apps to make fun. In the meantime being like in bomb shelters and being affected by missiles every day. That is why that like real heroes, anything that’s DNA of Ukraine like just finished, like speaking about the topic, I believe that Ukraine and those guys who horizontally says, horizontal grassroots initiatives that uniquely combined in the ways of the common enemy.

So when we understand that something is unfair, that we are gonna be. Affected our relatives will be affected. That’s the main intention and d n A of Ukrainians as a freedom and to be fair enough to defend our life and our will to be independent and to believe in normal life.

[00:47:37] Andrew Warner: And now you are so you, you are so committed to helping that you’re thinking about doing what now that you’re not in reface on a daily basis.

[00:47:49] Dima Shvets: I’m still advisor to

[00:47:51] Andrew Warner: Yep.

[00:47:52] Dima Shvets: to Cocos. So we I’m helping and with different strategies, pinpoints and different introductions and thinking of what’s next. Was the next, evolution of the company. But I believe that one of my. Intentions and roles will be, the intersection of Ukraines and the Western world because it happens that I have approach and access to a lot of investors, great people from yes.

UK, Europe. And that happens that when were started with phase one of one of the companies that supported a lot informational and technologically, And we have different interconnections in that. So I believe that I’ll definitely will support different DAO d military tech projects, different initiatives related to rebuilding Ukraine.

And that is the point where I see a lot of inner will to, yeah.

[00:48:46] Andrew Warner: All right, and so if I’m going to just take away a message from what I learned from you, it is to have big ideas, look for people who are executing well, and then go partner up with them without necessarily for you. Finding a problem that needs to be solved or being married to a specific approach. It’s more like I have a vision who out there can execute on in this world towards this vision?

I’m gonna partner with them.

[00:49:13] Dima Shvets: Andrew, correct. Just small example. Yeah. Like military projects I’m supporting now those guys had didn’t have a clear path or MVP or just understanding that it’s going to be a business. There is a necessity from general from Ukraine. Guys, we need help of this audio detection system. They did it.

They need some specific introduction strategy and. It was super complicated. If they were just thinking, oh, what about mvp? Going step by step, it wouldn’t even happen. But when they just executed, asked me, other guys helped it, and now everybody just thinking, okay guys, it wouldn’t, it couldn’t happen. There are so many restrictions, regulations, it couldn’t be true, but it happened.

So yeah, I think for now my approach is followed.

[00:49:57] Andrew Warner: Thanks so much for doing this, and thank you to lemon io slash mixergy for sponsoring. Thank you. Bye everyone.

[00:50:03] Dima Shvets: thank you so much.

Who should we feature on Mixergy? Let us know who you think would make a great interviewee.