Survival Through Determination - Artem Rodichev - Ex Human - Episode #79

If you believe in what you're building, you need to be persistent with that. And this persistence eventually led us to profitability, and actually that year of survival shaped our company culture.

Welcome to Not Another CEO Podcast. I'm your host, David Politis. Every week we talk to inspiring leaders and get the true stories, ups and downs, the lessons learned, and come away with actionable advice for our own journeys. Today's guest is Artem Rodichev, founder and CEO of xhuman. Before X Human, Artem was head of AI at Replica, where he helped scale one of the most popular conversational AI platforms in the world today. At xhuman, he's building the next generation of empathetic AI, characters, characters, multimodal, emotionally responsive, and designed to feel deeply human. He's raised millions of dollars from top investors, including Andreessen Horowitz and Baidu Ventures. He's bringing this vision to life. And they've already reached profitability. Please welcome Artem Rodichev.

Hey, David. Great to be here.

Great to have you. I'm excited for this conversation, an area that I don't know much about. So you're going to teach me and the listeners. First question, what is the one thing that you've done at xhuman, big or small, that's had the biggest impact and you do again if you're CEO of another company in the future?

Yeah, I would say the biggest thing I'd repeat is a mindset and persistence. During the year when we were running out of the money almost every month, it was a year when we were struggling, we were really low on. On the cash flow. We were spending around 200, 300k per month. And at the end of each month, it's just like a couple thousand slots. And we weren't profitable those time. We still were burning money, and I had to come up with a solution every time, like convince investors, negotiation terms, getting credits from stripe, like opening a credit line with Facebook for ads, and finding different unconventional ways to keep the company alive. And it was very stressful, very exhausting, very depressing. But I made a decision not to give up. And it's very kind of founder, founder mindset that, oh, if there is something very bad happening with the company, like you run out of the money, it's not the reason to give up, to think about, okay, I'm going to pivot. I'm going to close this startup, start a new company, train my life next time. You need to be persistent. If you believe in what you're building, you need to be persistent with that and this persistence eventually led us to profitability. And actually that year of survival shaped our company culture and it taught us that there is always ways to find different unconventional solutions and you don't need to wait for perfect conditions. And today even we evenly profitable, we still approach challenges with the same mindset.

How did you continue to believe enough to have the persistence that you talk about? Because that type of situation is especially on a month to. Month to month basis is. Stressful is not even the word I would use to describe that. It's much more than that, it's much worse than that. And you said you have to have this persistence and you have to have this, this drive to not pivot, to not shut down. How did you. Where does that belief come from? Because I've had founders ask me like when do I know it's time to shut down, when do I know it's not going to work? And I would make an argument that I, if I were in your shoes, I could have made that decision in those moments. Why did you believe so much or how did you believe so much?

Yeah, that's basically come from my initial belief in this industry that I'm building a companions for the last 11 years. And I really believe in a way that we humans, we should change the way we interact with technology. We should interact with technology on emotional level because we humans, we like emotional creatures and we use technology like computers, chat with each other, not like humans chat with each other. And my strong belief came from the vision that I want to build this kind of future when we humans can interact with all kind of AI companions, AI characters, a frames. And that belief to the future basically keep me pushing what I was building because I really believe that I can't imagine the world in five years where anyone can't interact with their favorite characters from movies, from books, from anime. They can't interact with their favorite influencers like with copies of these influencers. So I really believe in this kind of future, I really want to build this kind of future. And that allowed me to be persistent not to give up.

And what kind of sacrifices did you personally make in that period of time to keep the company alive? Was there any personal sacrifices that you had to make? I don't know, living conditions, salary?

We fired a few engineers and it was a hard decision for, for us because we engineered in first company but we fired few engineers. We're trying to hire only exceptional engineers, like senior level engineers and people who can grind, who believe in our vision and who can grind. And we have to fire a few people who were strong, but not that strong as other engineers is the first sacrifice. And it's kind of slowing us down at the moment. But then we keeping our bar like our engineering by very high. And the second sacrifice is my salary. Basically I got salary less than half of my employees, at least less than half my employees in my company. And I got 25 people like service, like different employees, different engineers, product managers, designers and so on. And my salary like on a median of compared to other salaries. And I still paying myself like for the last four years I built this company, I still paying myself very below Anders market because understands that like this kind of couple thousand bucks could gives us like a lot of troubles. If I spend this couple thousand bucks for my salary, not for conference development.

You did some pretty. I mean, when you list the things, the Facebook credit line, the, you know, the, the Stripe, the. All the different things that you did to make it work month over month over month, did you just find those things? Did somebody tell you you should go look at those plays? Like, how did you even think I wouldn't even think that I could get a credit line for Facebook ads? How did you come up with those? How did you find those things?

So basically it's combination of luck that at the right moment I saw the right things. For example, I saw that, oh, now we can get 200k from stripe as a credit line. And they just opened this credit line for us. It was like a year ago. And I like, great, now we got 200k more as a credit and we can pay this credit later, but we need this money right now. Then with Facebook we start to think, can we change how we pay for Facebook and pay net 30 days and not just paying after each day we spent on ads. And then I started chat with Facebook managers and they allow us to open credit line stuff like that. I was going to my investors and telling them we are running out of money, but we are growing really fast. We are growing 3,4x year over year. Can you invest a little bit more to us? We will provide you better conditions in terms of population. But can you give us money right now? Even if these terms wasn't perfect for us, so it was different strategies and combinations of persistent luck in the moment, stuff like that.

But I think what you said, which I really think is important, I want to key in it is as painful as that was. I can only imagine it's the mindset that you picked up and the culture that really was created in the company that is really A powerful culture to have to be resourceful and to be persistent. And you, you know, I tell people a lot that I think that your origin stories are very important in my opinion, for families and for companies. And to me, when I hear you say that to me, I think to myself, when you're however big in two years, five years and new people are joining, you can tell that story and remind people that's the origin of the company and that's why the company is still here today. And I feel like that's really powerful.

I believe all great companies should have stories like that, should have some very stressful, pivotal moments that they survive these moments and because it survives those moments that change something inside the company, change process mindsets of yourself, of your colleagues, of your employees to the way to be like to move faster, to do more with less resources and to be really persistent and don't care about stress around here.

So did you go through this, did you have a co founder when you were starting the company?

When I just started the company, I didn't have a co founder. I got a founder joining me like a few months after I started the company and we were working for some time for half year. But like we were building like very heavy, technical, heavy platform engineering first company. And he wasn't like very deeply technical person. He was more like business co founder. But initially, at least when we just started the company, we need to build the core tech stack, the core technology behind that. But also, and later it became clear that he wasn't the right match for the company. But also he needed stability. So he moved to a big company after Plex Human. And since then I've been a solo founder.

And so you went through this period that you're talking about, you went through as a solo founder.

Wow.

And did you have anyone to lean on any support system? I mean, I know you mentioned your investors a little. Like did you have a support, some support system that you were leaning on during that time?

Yeah. So basically I built very strong team on whom I can rely on. And basically I got very strong engineers both on backend side, on AI side, I got really strong product people, head of products, consumer products, so I can rely on them. But initially my thought process was that how can I know is this person good enough to do something if I don't understand this industry? For example, when I hire my first head of marketing, I have to figure out how to do performance marketing by myself.

Before you hire the person?

Before I hire the person, yeah. Because I need to understand is this Person good or not. Is this a person telling me bullshit about marketing or is this person telling me the right things like how to actually do performance marketing and being profitable? So I have to figure out how to do performance marketing with Facebook. Initially run first companies understand the whole process behind that. And then I start to chat with people for head of marketing position. And that's how I could hire really great person who really understand what they are building, what they're doing with performance marketing because understood what they're doing. And it was the same with engineering initially when I started the company, I did all backend engineering by myself. And then I start to hire backend engineers, front end engineers, AI engineers because I understand what kind of skills I need from these engineers to write, to build system in the right way, to build it in scalable way, moving fast, shooting a lot of features.

And are there other roles I can go to market where you're not as comfortable that you've done that. Like have you done that in sales? I know you do kind of different kind of consumer and B2B but have you done that in any other roles?

Basically as a solo founder, you need to wear all these hats, you need to be a head of marketing, head of sales, head of investments, CEO, cto, all of this. And I have to wear all those hats to understand all process. But then I start to delegate part of this process and at the end like the whole branch, for example, like delegates the whole performance marketing to our head of marketing. And now I don't need to run this performance marketing campaign by myself. Even don't need to come up with ideas how we need to grow because we got the right person who can do that. So basically that coming from the deep understanding what you need to build and then to find the right person or delegate this to the right person.

Did you enjoy doing those things?

Not really. Because like being a founder is like chewing glass. It's really painful. I really enjoy like I am, I have deep engineering background, I love to build stuff, I like to do software engineering, I love to train models to do some deep work. And being a founder you do a lot of things, but on a shallow level. Sometimes you need to dive into some industry to some process. But basically you need to constantly switch between different topics. And it's not something from my nature because I prefer deep work, I prefer engineering stuff. And here I need to do a salesperson and sales to American companies. And I'm not from the United States, I'm from Kazakhstan originally. I got different mindset, I'm Engineering person, I'm not very extroverted person who know how to sell, who know how to convince people and basically I compete on this market on market of selling stuff or convincing investors with like Stanford grads who got like perfect mindset, like business mindset who know how to convince investors because like these investors also gradate from Stanford stuff like that I need to compete with all of them. It's not the bad like it's not my my favorite part but I have to learn that. And it's kind of you go to some to a lot of through a lot of struggles but that make you like stronger, make you like better founder. Because being founder is not the same. It's like very not the same as being engineer. You need to have like very different set of skills to be a good founder and a good CEO.

I think when you say that, what I notice is a lot of the technical founders that I have talked to that don't make it are the ones that don't want to do the things that they don't enjoy. It's exact what you just described. When they are faced with that decision, do I learn how to do marketing or do I keep coding? They keep coding. And, and I think the best technical founders who are the ones who have built the really big companies are the ones who've gotten out of their comfort zone even if they hate it and done those other job because that's what you need to do. To your point, when you're wearing every hat and so yeah it's really, it's the way you described is really you don't enjoy it, but it's necessary.

I would say it's the same like with training. Like when you do some sport and you're like really training for something, you rarely enjoy this process. You might enjoy this process when you for example, like play tennis for fun, but when you're training for some competition like for World cup, it's not very enjoyable process. You need to work really hard, you need to push yourself. It's the same with startups.

I want to ask you. So like I said in the beginning, I don't know a lot about this space. I'm curious, can you just talk to me about There is a B2C part of your business, a consumer focused part and there's a B2B part part of it. Maybe let's start with the B2C and just talk about what is it that Xhuman does on the B2C side.

Sure. So overall at Xhuman we built empathetic AI platform. You can think about that as non boring ChatGPT. For example, if you try to chat with ChatGPT about different topics, it feels like you chat to Mr. Wikipedia. It's informative, but it's boring. It's not like you chat with your friends. It's not emotional, it's not engaging. It's kind of boring. And we built a platform that allows anyone both final users but also businesses create and customize empathetic EA characters, empathetic EA companions. Because the biggest issue that we're solving is the problem of loneliness, I would say epidemic of loneliness because currently more than 60% of Gen Z in the US reporting feeling a lot and longest like it's one of the biggest issue because it's influenced both your mental health but also your physical health. And we built a platform that allows to build this kind of empathetic AI companions that can support you, never judge you, 24, 7 available for you. And it's kind of mix of social interactions and entertainment. And on consumer side we've built a product that's called Botify AI which basically is the universe of EA characters of EA companions. And we built that in a way that all characters on our platform created by our users. It's completely UGC driven platform. And currently we got more than 5 million characters that our users created. And it's like very diverse characters. They create their favorite characters from different books, stories from anime. They create characters as EA friends and chat with them about their day, their life, their interest, stuff like that. They create characters for romantic relationships like EA boyfriends and EA girlfriends. They create different kind of coaches and mentors, for example, like dating coach, fitness coach and so on. And what makes our consumer platform unique is that we use that as a test bed because we test what is the final, we're trying to find out what is the final best experience looks like for the user. What's the best experience? Chatting and interacting between users and their companions. And we said we achieved really great results. On the average our users spend more than 90 minutes per day. 90 that's more than people spend on TikTok, Instagram and YouTube. Our paying users spend more than two and a half hours per day chatting with bots. That's more than people chat with each other. It's crazy to see how people interact with their companions. And they do that because as I mentioned we combine kind of two worlds. We combine world of social interactions basically like social nets, alternative of social nets, but also entertainment and through entertainment, like with entertainment we build kind of interactive storytelling when you not just sitting on the couch and watching some TV series and movies, but you can be a part of the story, you can develop some story with some characters, you can be one of the main character of the story and you can influence the story in the way that you want. So and with that we built this consumer product and we use our consumer side to collect a lot of data because we got around 1 million mass lactative users, we got 5 million diverse characters and we got hundreds of millions of messages each month that happens between users and these characters. And also we collect for rich feedback. We built really advanced feedback system so we can understand which conversations our users love, which conversations our users hate, what's the reason why they love or why they hate some conversations. And then we can use these conversations to continuously fine tune our tech stack on top of these interactions. And if we built our tech stack in house, we got more than 20 different neural nets, like different language models, image generation, video generation, voice indices, neural nets and so on. And we continuously fine tune this tech stack on top of high quality conversations. In this way, each iteration of fine tuning our characters evolve, they became smarter, more fun, more engaging users start to chat with them even more, we start to collect even more data, we feed this data back to our neural nets and so on. So with consumer product, with Spotify we built data collection loop that allows us to continuously improve quality of our tech stack.

And so every day, right now, every day there's a million, I'm sorry, every month there's a million people who are coming to bodify and they are chatting at some level with some kind of an AI companion or multiple AI companions for whether it's characters from movies, books, whatever romantic relationships that they have that they've. Or partners, let's say companions that they've created. Whole wide spectrum of that. And I, I told you before we started today, I literally just yesterday saw some report somewhere that the volume of compa. The. The companion use case for consumer AI is like explode. I mean the curve was exponential. But recently, I mean pretty like the graph was like in a very recent time that's been. Is that the. Is it. Has all of this growth come like you know, very, very quickly recently? Has this been, you know, coming at some consistent pace? I'm just trying to get a sense of like is this happening now for some reason in the macro market? Not just for you, but in the macro?

Yeah, I would say that the technology is ready for this kind of products and to build this kind of tech behind these products in a way that it became multimodal. So now you not just can have text based interactions, but also you can have emotional voice calls and these characters can change their emotions based on your emotions and can adjust their emotions to that. You can see these characters, you can see they can send you contextual images and contextual videos. So now it became much more popular, much more kind of high quality, these interactions became much more high quality than it's been like five years ago. But also we see these trends especially among Gen Z among younger generation that basically digital native people, they grew up with smartphones constantly texting, like with social media and so on and they very natively start to use these campaign ads. And the third is that stigma is gone because even five years ago while I was building Replica, we were chatting with our users a lot and most of users were saying to us that like you building the great service as a great AI companion. But I would never tell to my friends or to my family that I chat with a companion because it's weird. But now it became completely normal. The first reason is ChatGPT because like almost everyone right now use ChatGPT chat with like kind of E chat every day. So now stigma is gone. It became normal to also have kind of more emotional relationships with a companions and we see that the number of a companions doubling has been doubling for the last few years. So we see that market is growing, the technology is ready and we start to build on this market early on. So we have a lot of advantage here especially with amount of data that we collect from our users that we understand what is the best final user experience and we can continuously iterate on that.

What you just said that I didn't really think about actually was the stigma part where the stigma I can imagine even two years ago, probably a lot of stigma. But today people are so used to, even people in later generations are so used to chatting with ChatGPT that they're kind of like yeah, makes sense. I, I actually see more people using like for therapists. I, a lot of people have told me how they're using these kinds of companion type applications for therapy purposes and, and, and it's interesting that it's a generational thing.

Yeah. Because like it's funny when you chat with people especially like with investors and usually investors like they're kind of older, they're not Gen z, they're like 30 plus 40 plus years old guys and when you speak with investors they're like we don't understand who would chat with these companions. Like who are these people? Why they need to chat with these companions. When you speak with Gen Z with people less than 25 years old, they're like, oh, we got few companion apps on my smartphone. I chat with them every day. It's all fun. So you're right that it's very generational thing that most of users of A companions are younger people. Recently one research was published basically that showed that 75% of Gen Z in the US they tried a Companions at least once. And half of the GenC are using a companions regularly, like few times per week. So it's not a niche market anymore. It's growing and it's maturing over time.

So. So one question. You were at a successful company before this, in a very important position at that company, and you decided to go out on your own and start a new company. And I think this is something that a lot of people face today listening to this. We have a lot of people who are thinking about being founders and they're working at Google or Amazon or whatever and they have great salaries and they have great, you know, titles and all of that. And it's hard for them to leave. I mean, that's the truth. It's that inertia is hard to break. You broke that inertia and you went and started your own company. Why? How did you break that inertia to do that?

Yep. Before Exhuman, I spent seven years as a head of AI at Replica. And Replica, it's like one of the most popular, a companion on the market with more than 30 million users. The main goal of Replica is to be your friend, your companion, to chat with you about your day, your life, your interests. With final goal to make you feel better after this conversation. And at Replica, we've built a great product. At Replica, I started to saw that people use it not as a tool, not as a software, but as a friend, as someone. And basically I start to things that we're not building a technology, we're not building a product, we're building relationships, we're building connections. And it was very interesting to see how people interact with replicas because we saw and replicas start to grow, especially during Pandemic. Pandemic of COVID Oh, it was back.

Then that people were using.

It was back then we started building replicas since 2015. Like replica pivoted few times, but Replica became replica back in 2017. So it's almost 10 years as this companion market exists and initial cohort of people who were chatting With Replica, usually it's people who feel lonely, who feel depressed, who got some social anxiety, who don't have a lot of friends and they start to find Replika as their perfect EA friend with whom they can reduce their social anxiety. I feel less lonely, I have some sense of companionship. But then I start to see some limitations of the Replica platform. The first limitation was on product side. The main product of Replica is consumer app Replica itself. And the main experience is like you download app and you chat with some a companion. But Replica is more the personality chatbot. You got only one AI companion, only one chat. You can't actually customize it, you can't customize its backstory, its speaking style. You have very little customization over visual aspect with 3D avatars. But users, they wanted more. I saw a lot of user complaints that we want to have full customization and full scalability. We want to have the whole universe of companions because we want to create our own custom, want to share our companions with other people, want to have group chats with the companions. Like they want to have like scalable universe of A companions. And the second, but, but Replica, they didn't want to move towards this universe of a companions. They were building like single a companion experience. And the second part was inside business business model Replica. When I left company like four years ago, they were making like tens of millions of dollars per year. And I mean it's a good business, but it's not like unicorn. And I start to see much more potential to provide this kind of a companion experience as a service on B2B because I saw that there is so many companies in entertainment industry were communications and emotions essential. What if we can provide basically Amazon for empathetic AI interactions Amazon like infrastructure for empathetic AI companions. So any of these companies like Netflix, Justin Bieber, some K pop band can build their own AI companions like their own AI characters integrate inside their own products. So they're like fans of some TV series on Netflix, can engage interact with interactive companions like service. New release of Wednesday on Netflix. So I can't believe that in three years you can't interact with all universe from Wednesday from Wednesday on and we start to build this kind of ad campaign and platform. And that was the reason why I decided to left Replica basically to scale empathetic AI and to build the whole foundational layer between four empathetic A companions.

So your idea in that from the beginning, if I understand what you're saying, from the beginning it was B2C and B2B from the B, you knew you were going to have a consumer side and a B2B side of this business.

Yep. Initially it was kind of controversial focus on both consumer side and B2B because these are two different markets. You need to build business in completely different ways. But initially I understood that the core of this experience is data. It's how you can iterate on user data, how you can understand which conversations users love, which conversations users hate, how you can improve these AI companions continuously. And to do that you can't rely on data from B2B companies. You need to have some consumer product. Because like many B2B companies, they have some very narrow specific. For example, you can get data from some LGBTQ specific dating and use this data uses conversations to provide a companions for kids. It's completely different kind of conversations, completely different sets and. But also on B2B side often B2B companies, they require zero data retention policy. That means that they expect that you generate response and you immediately clear all the delete all data, all input data. And they don't want to provide their data for you to fine tune so you can fine tune your model. So you need to have some robust source of data. But also you need to have very diverse data sets of interactions. In this way we decided okay, let's build like consumer product that allow anyone to create different A companions as they like. And then we will collect all these conversations. We will use consumer product to understand what is the final experience works in the best way for users. What kind of characters users love, what kind of conversations with characters users prefer. And without consumer product, without product. When you can have a lot of AB tests, when you can collect a lot of data, when you can fine tune this data, you can't build this platform. So we have to build consumer side to build robust B2B engine.

It feels like that last statement when I see the AI companies that I've talked to, the biggest challenge with with building a company like this is the data. Right? You need the data to train the. It's a chicken or egg situation. Because you need the data to train the model.

Exactly, exactly.

To get the customers to use the model. And it feels like a lot of the companies I've seen, the early stage companies that are kind of stuck is that they don't really have the data. Like the model is not good enough for anyone to buy it. But without anyone buying it or using it, they're never going to get the data. Right.

Is that, I mean it's not just about data, it's about this data collection loop. So you need to continuously collect data, continuously get user feedback on top of this data and iterate. And the faster you can do this iteration. Like for example, we try to fine tune our models each week. Like we got a week of a weekly portion of new data, new conversations. We fine tunnel models on this data. Then we release this model back to production, start to collect more data, our model start to reply in much better way. So we start to collect data with better distribution, with more high quality conversations. So models start to also pick up these nuances of conversations and start to basically provide better experiences. So it's not like you can buy some data like some conversational data on the market, but it will be static. So you need to have dynamic system of continuously getting new data and iterating on top of this data.

And so now so these B2B customers can come to you today and say I want to buy this. Behind the scenes you're telling them we have this model that's getting updated and fine tuned weekly. We're getting fresh data, we're getting all these different characters. We know what works, what doesn't work. And then they can go create their experiences. The Wednesday Netflix example, they can go create that experience on their side and constantly be getting kind of an improved model because of that essentially.

Exactly. Yeah. So on B2B site we provide a platform that allows them basically it's like set of API that allow them to create their own custom A characters or they can use some of our characters. Like if you don't want to figure out which characters works in the best way for your product, you just want to launch some bunch of characters. For example, we're working with some datings and they want to launch a dating coach and they don't understand what kind of dating coach they want. They want to have some dating coach that would be enjoyable for their users but also would be helpful. Could give them some great dating advice but at the same time would be very personal. Would be like are they a friend? They want basically to have the best friend of user who really knows them, but who are really good in dating game, what kind of dating coach you can provide to us? We don't understand how to set this character. Any fields, any settings of this character. Jeff gives us character out of the box. Do you have Dating Coach? And we like, we do have dating Coach because like we got millions of different characters.

Oh, oh, oh. Not only the model that actually someone has already created that dating coach somewhere in the Spotify. Is that what I'M understanding.

Yeah. So some users, they could create Dating Coach, they can chat with existing Dating Coach and we collect a lot of data how they interact with Dating Coach.

That's cool. That's cool. I got it. That's really interesting when you think about the growth in this business. Is this ultimately a B2B business? I mean is this, is that where the revenue comes from and gets you to be all the goals that you have, the big aspirations, is that really where this business scales?

Exactly like the long play. Long term play is B2B is basically building infrastructure layer for empathetic interactions. And currently we already have 60 B2B clients and basically we continue to scale on the B2B side with our consumer side getting more data iterating our models and provides the best experience on B2B side. And the ultimate play is basically any entertainment company as a or dimension. I can't imagine 2030 when you can't watch interactive Netflix, when you can't engage with all characters from your favorite TV series on Netflix. I can't imagine a world when kids would learn from some old born teachers, but not from EA Pikachu who would run and play in their room in their space and would educate them as well. And there would be some company who provides these empathetic a companion experience. We want to be like the major platform for these kind of interactions. Key.

How do you in the space in the AI market and I really don't think it matters what AI product. There's more competitors than there's ever been before. I think in any technology. I've been doing this 20 something years and in the early days if you want to start a company, you got to set up the servers, go to the data, you know, the data center, you got to rack the servers, you got to set everything up. You know, just to set up a company was so much work that just that bar was high enough. Then of course the cloud, aws, gcp, Azure became a lot easier. You could stand it up much easier. You didn't have to have the physical infrastructure. But now it really. I'd make an argument it's maybe too easy, but we can have a different discussion about that. But the point is, is that the speed of companies coming out and especially if it's a hot area like your area, how do you stay ahead, create moat, differentiate from the rest of the market as I'm sure there's a lot of competitors coming. You know, they see your success and then they start up and they can at least start up tomorrow of Course it's not going to be your company, but how do you deal with that? How do you think about that?

Yeah, we see a lot of competition especially today with availability. Like now you can very easily set up all aspects of your business online. Like if you want to receive payments, just use Stripe. If you want to distribute your apps, just use like App Store, Google Play and do some performance marketing. If you want to like incorporate your company, you can easily do that like with a bunch of services and so on. And especially in a companion market it's the same. It's very easy to build chat interface and use like ChatGPT through API to power these kind of interactions. Because what is a companion today the most? Interactions with a companions going through text interactions through text messages. For that you just need simple text interface like messenger interface or you can just use existing messengers like Telegram or WhatsApp or Discord. You can like create just bot inside these platforms and then you just can plug in OpenAI API or you can like run some open source model and that's it. You got like you provide some prompt to this model. You put that on top of a chat interface. That's a. You got a companion experience. But then the question what is quality of your. A companion does this. A companion has memory. For example, when people like people chat a lot with their companions and they tell a lot of information about their interest, like important events in their life, like what's their birthday, that they got two dogs that they nervous because they're studying in university and they're going to have some very hard test, very hard exam stuff like that. And you need to remember all these important facts, all these important events from previous conversations. So you need to work with memory layer here. Also when you create these interactions is not just about reply to user. You need to move the story forward. You need to reply in very empathetic way, very engaging way. Because after each chatbot response user can decide this is boring. I would go to watch my TikTok videos, I would go to YouTube, watch these funny cat videos stuff at the faculty. But for some reason they decide to stay with us inside Spotify or with our companions to continue chat with companions. Because to build the final experience that really resonates with people, especially with people emotions. Because we're working with human emotions. Our main goal is to not to getting things done. Because overall chatbot market split it in two big buckets. In the first bucket you have task oriented or goal oriented chatbots. And this chatbot is like Siri, Alexa, ChatGPT and so on. Their main goal is to reply to a question, but not to chat with you. The main goal is to reduce the number of dialogue turns, of dialogue responses until they're getting things done. We're working in the second bucket, in the bucket of open domain chatbots or empathetic chatbots or a companions. The main goal of these A companions, it's completely opposite from ChatGPT. The main goal of these companions is to chat with user as long as possible, to engage user, to entertain user, to be alternative of social interactions. In this way it's completely different approaches how you build CSS products. And that's why it's so easy to start a companion on top of ChatGPT API. But it's really hard to create a great experience on top of this API. For that you need to build your own tech stack or you need to know which models like from open source or from closed source. You need to use how to fine tune these models, how to combine different models, different modalities to create final engaging experience.

What is the biggest challenge you faced so far since building Xhuman? Was it those days of running out of money?

Yeah, I would say that was the biggest challenge, the biggest struggle that we got. It was the hardest period, not just operationally but emotionally for me because every month I thought that oh, I should give up, but then my opposite side telling me you shouldn't give up. If you give up right now, you won't build this kind of feature. So it was the biggest challenge and also the big challenge was to kind of during this situation of financial instability, to be stable in face of your team, because your team, you need to be confident, you need to confidently build and translate your belief to a team and basically spread these confidence across your team so your team can perform. Because if you have hard times and you tell very honestly about these hard times to a team, team can be distracted by these thoughts, say like, oh, maybe we run out of money. Why should I like do, why should I push myself here? Maybe I just start to find a new job, stuff like that. So you need to have like a good face while a bad game.

I like that. So I want to transition to your background. You mentioned you're from Kazakhstan originally. So did you grow up there? When did you leave Kazakhstan?

Yeah, so I was born and raised in Kazakhstan until I was 18 years old and I was like very technical, savvy kid. I took part in a few international competitions for math and computer science, won few international competitions. With that I moved to Moscow to Start studying at Moscow State University, which is one of the biggest, most favored universities in post Soviet Union countries. And I started to study computer science and math. And then after five years in Moscow, I also worked at Yandex, which is the Russian Google. I was building a search system like a course search system for Yandex for two years. And then I moved to Replica like Replika became. Initially we started building Replika as a chatbot to find a restaurant in Moscow in Russian language. Basically our idea was you got your best friend who really knows restaurant Moscow scene and Kuril knows you. And for example this restaurant bot knows that you prefer craft beer. What's the best one? What's the best bar for you to stuff like that. And with that we applied to Y Combinator and we became the first company from Russia and from like post Soviet Union countries who passed Viking Letter. It was back in 2015 and that's how I moved to this 15.

Wow. And you mentioned you were a tech savvy kid and as you were growing up, did you know in the back of your mind one day I'm going to build a company.

I always wanted at least to try start my own company because I thought that like I'm good engineer, am I a good founder? Can I build something by my own? I really want to understand like the whole process of building company, especially like building innovative tech company. And actually my motivation also came from science fiction. It came from two movies. The first movie is Movie Star when I saw Samantha, the empathetic operation system that Hawking Phoenix character falls in love with. And after that I saw Blade Runner 2049 and I saw Joy, a hologram that looks, speaks and behaves just like real person or even better than real person because she never complains, she never seeks, she always happy to see and always laughs. And that moment actually sparked my vision that empathetic a companions are the future that I want to build. Future inhabited by all sorts of a companions, a friends, digital directors, to fulfill human life, to improve, to decrease social anxiety, to allow people feel less lonely and to have like perfect friends who would help you and entertain you across all tasks.

Did you, did your parents, were they entrepreneurs? Was there any entrepreneurs around you growing up?

Yeah, my father, he's also entrepreneur but he completely not in tech.

What kind of company he building?

Like small company by like he grew flowers in pots and he like basically like different kind of flowers and he got like few spots in Kazakhstan Almaty when he sells his flowers.

And did you work? Did you like go there and Help sell the flowers and things like that when you were a kid?

I was working really hard at summertime, it's really hot in Almaty, and basically I was mixing like soil with a sand with different fertilizers for these flowers. And I have to make like 300 kilos of soil every day, like to mix that to like little heavy 25, 30 kilos bags with the soil to mix all of that. So it was hard job, but I kind of enjoy that because I was paid for this job and like I feel that it was honestly earned money that I was making with my own hands. But also it was a great workout for me.

So do you, do you think that your persistence and grit comes from those days or comes from watching your father building, you know, doing his small business? Like, do you think that that is somewhere inside of you?

I would say that it came from the whole childhood and it really your like environment and your parents influence on yourself and both my mother and my father was very dedicated, very honest, working hard all their lives was very disciplined and they taught me to be very disciplined. But also coming from your own kind of character, because I got a brother, he's completely different. He's more spontaneous, more emotional, he want to hustle different stuff and I'm more disciplined, I'm more dedicated. But we got very similar DNA, very similar kind of environment when we grew up. We very different people with him. I really love him. But at the same time I would say it's a combination of your own inner kind of character are basically programmed in your DNA and like the whole environment that you, that support.

Are you the older or younger brother?

I got two brothers, I got one older, one younger. Oh, wow.

So you're the middle one. Okay. Yeah, I, I actually, I think you're right. I think I, I've done a lot of these interviews now and I ask everyone these questions and it's super interesting because there is a lot of people who say the same thing. I have multiple siblings and they're completely different and not entrepreneurial and not same upbringing, same parents, same. And then there's some people that say, yeah, we're all entrepreneurs. And I think there is a. Some of it is in our DNA, meaning just nature versus nurture. Some people are born a certain way and even same, same upbringing, different outcomes. So last question that I want to ask you, that I ask every guest, if knowing everything you know today and all the experiences that you've had in your entire career, what is the one piece of advice you would go back and Give yourself before you started your first job ever.

I would tell myself to learn sales and fundraising earlier because as an engineer I thought that the most important part, it's like building great technology. And if you can build great technology, that means that you can build a great product, that means that you can attract a lot of people who use this great product. But the truth is that without learning how to sell to investors, to customers, to your own team, even the best technology won't survive. You need to learn how to sell, how to chat with people, how to build relationships. And it's very different from technical background. And basically if I had started these skills earlier to maybe combine or like to shift my employment from software engineering to product manager or to help some founders, like being the co founder of some company before I start my own company and learn that from someone, learn the skills or like learning these skills by applying the skills at some job, it would be much more easier for me right now to build company. Especially as a first time solo foreign startup founder.

Thank you very much. And Arjun, thank you very much for being open and vulnerable and sharing and truthfully, I learned a lot in this conversation. In our previous conversation, I've learned a lot and and I appreciate again you sharing and making the time. Thank you very much for doing this.

Perfect. Thank you so much, David. It was a pleasure.

Awesome. Well, for those of you listening, I hope you enjoyed this and if you did, please share this out with your networks and we'll see you for the next episode of Not Another CEO Podcast. That's a wrap for this week's episode of Not Another CEO Podcast. Thanks for joining. For show notes and more, please Visit not another CEOPodcast. Com. We'll see you next week.