How Praktika Built a Profitable $50M ARR AI Business — Adam Turaev, Praktika

David Barnard:
Welcome to the Sub Club Podcast, a show dedicated to the best practices for building and growing app businesses. We sit down with the entrepreneurs, investors, and builders behind the most successful apps in the world to learn from their successes and failures.
Sub Club is brought to you by RevenueCat. Thousands of the world's best apps trust RevenueCat to power in-app purchases, manage customers, and grow revenue across iOS, Android, and the web. You can learn more at revenuecat.com. Let's get into the show.
Hello, I'm your host, David Barnard, and with me today, RevenueCat CEO, Jacob Eiting. Our guest today is Adam Turaev, co-founder and CEO of Praktika, an AI language tutor app with more than 30 million learners. On the podcast, we talk with Adam about the month they went from a million to 12 million in ARR, selling Brazil on career upgrades instead of English lessons, and why you shouldn't tell influencers how to do their job.
Hey, Adam, thanks so much for joining us on the podcast today.

Adam Turaev:
Sure. Yeah. Hey, David. Hey, Jacob. Very nice seeing you.

David Barnard:
And Jacob, always nice to have you on.

Jacob Eiting:
I'm excited to talk about language learning and apps. Two of my favorite categories of things.

David Barnard:
Yeah, we were just talking before the podcast. You're a longtime language learner, so this is right up your alley.

Jacob Eiting:
I've been on a break. I went on a break, but yeah -

David Barnard:
To be CEO of RevenueCat, but fair enough.

Jacob Eiting:
A little bit, yeah. Yeah, I've been a little busy with that. But yeah, definitely I like to learn languages as a fun party trick.

David Barnard:
So Adam, I wanted to kick things off talking about the pivot. So it wasn't a pivot in that you pivoted inside the same company, but you were doing enterprise chatbots with Cleverbots. Why did you switch into consumer language learning? And then especially, why did you switch into consumer language learning when Duolingo is like a 900 pound gorilla and everybody's like, "Oh, don't compete with them."

Adam Turaev:
To explain this pivot, I would say, so together with my co-founders, Anton and Ilya, we're schoolmates actually, and we've been building projects together and always dreaming about building something innovative and giving value to people worldwide. And there are so many pain points that you can resolve. And we live personally through language pain point, I would say, and language barrier pain point to be more precise.
At the same time, we started with conversational AI. We really dive deep into this sector and started thinking about, all right, how can we leverage conversational AI and give value to enterprise customers and build our first company called Cleverbots? It was back in 2017, so long before LLM era.

Jacob Eiting:
Same year the Transformer paper came out.

Adam Turaev:
Yeah, yeah, yeah. Back then LLMs were called natural language processing, NLPs, and we actually saw a lot of potential in this. And I think it's just a little bit of an off-topic here, but one of the first inspirations for conversational AI was actually by Telegram's Pavel Durov, when they launched API for chatbots in Telegram. And that's what inspired us, by the way.

Jacob Eiting:
Yeah, there was a pre-LLM wave of chatbots in the mid-teens.

Adam Turaev:
Exactly.

Jacob Eiting:
Facebook was pushing them. Everybody thought it was imminent and then it never developed.

Adam Turaev:
Exactly. So we got super inspired by this and our longer term vision was, "Hey, that's how people are going to interact with machines."
And so we started digging into this and started seeing actual value we were giving through these chatbots and voice was to enterprise custfomers who were working with, largest companies, worldwide, international companies, and we automated their customer support, HR support, a lot of tickets, sometimes up to 60% of tickets imagine at its pre-LLM era.
But the more we work with that, the more we realized that conversational AI can be actually used to resolve our main pain point, which was our own limitation in terms of what we can achieve if we break through this language barrier issue. So I'm obviously not a native English speaker and English is definitely an income multiplier, opportunities multiplier. So we just realized conversational AI could be used for that, and we saw much more potential there.
And again, it was pre-LLM decision. That's how we pivoted from Cleverbots to Praktika.

David Barnard:
It's always fun to hear a story where you were right, but the timing was a little off because had you started Cleverbots maybe 2020 or 2021-ish and been ramping up when LLMs hit, you had been really well-positioned there. But instead, what year did you pivot to Praktika? Because I imagine that it's kind of a similar story, though. It's you transitioned into Praktika and then you rode the LLM wave, but just on language learning instead of enterprise chatbots.

Adam Turaev:
Yeah. Publicly you'll see that we launched Praktika in 2024, but this journey is much longer, to be honest, because we started thinking about this idea and how we can use conversational AI back in 2021. So we started building first prototypes, they were pretty bad.
Then every year as you know -

Jacob Eiting:
I was going to say '21, when did the first GPT API come out? That was '22, right?

Adam Turaev:
I think probably around same time.

Jacob Eiting:
But they were so broken though. It was like you couldn't really have a conversation with it.

Adam Turaev:
Exactly. But I remember us using their API back then.

Jacob Eiting:
Could you tell? Did you just have conviction like, "Okay, this is going to scale, this is going to get better?"

Adam Turaev:
100%. It's easy to say now, but 100% conviction.

Jacob Eiting:
Well, if it went the other way, you just wouldn't be on a podcast right now. That is the difference.

David Barnard:
Survivor bias for sure.

Jacob Eiting:
Yeah. I think I was just tweeting about this yesterday because I think I've underestimated these things the whole time. I've always said they're interesting novelties and all this stuff.
But obviously for the folks that did believe before it was mainstream, but you had to have a lot of conviction to say, "We're going to stick this out. We're going to invest in this even though the models aren't there yet. They'll get there, we believe."

Adam Turaev:
Yeah. And especially when everybody's going crazy about Web3 back then.

Jacob Eiting:
That was not the time to do an AI product.

Adam Turaev:
We were with our AI, "Hey guys, we're automating with chatbots."
And it was actually a lot of promise around chatbots. And those chatbots that we were selling, the expectation was of current chatbots. So that's why the time was a little bit off. But we always knew it's just a matter of time, technology will ramp up.
So I think this early belief just allowed us to be one of the first to tackle this specific pain point.

Jacob Eiting:
Also just to get good at building with these things, right?

Adam Turaev:
Yeah, exactly.

Jacob Eiting:
I think that's something we've learned this year at RevenueCat is we were behind and we weren't forced to learn these tools and now we're catching up with agentic engineering and working with these APIs and all of that stuff, which I think is an advantage if you've been working with them for years.

Adam Turaev:
Yeah. I mean, there is even specific sector, I would say, conversational user experiences and the way people expect chatbot to react, the way they expect it to respond, all of these moments when they churn as well. We knew about those basic moments, building hundreds of those chatbots pre-LM era. So yeah, I would say it allowed us to be pretty opinionated about what we're building. So I think that was the point of it. Yeah.

David Barnard:
I'm curious with both Cleverbots and Praktika, did you raise money? And I'm always fascinated if you're willing to dive a little deeper, the whole idea of pivoting, whether it's from one company to a completely new company or pivoting inside a company, I'm always fascinated to hear the kind of nitty-gritty behind that because I think there's so many people who should probably pivot, either things just aren't working, there's more opportunity.
And so getting into the nitty-gritty of how you made that decision, and then what the transition was like, shutting down or putting on ice the other company, starting the new company, did you raise from the same investors or did you not raise? I'd love for you to dig into that a little bit.

Adam Turaev:
Look, I think it was really decent desire to resolve this pain point and knowing that conversational AI can tackle this, not us chasing, all right, where's the biggest tongue? Where's the biggest market? Et cetera. So it was really our understanding that there is a pain point. There is, I would say, a dissatisfaction with current solutions. It was really sad to see that nobody's doing that. When we realized there are no solutions like that, we just went in, we knew that there is a huge market. We didn't expect it to be that big, but yeah.

David Barnard:
And your focus is more career English for job seekers, rather than a Duolingo that's more casual and more of edutainment. And it's such a classic kind of differentiation story that you went into what a lot of people would never... A crowded market a lot of people would say you should never enter, but you came to it with differentiation. So I'm curious the thought process behind that.

Adam Turaev:
Yeah. But again, even with Duolingo, I think, again, they built an amazing product. All kudos, numbers are still there.

Jacob Eiting:
Maybe the first and only large scale education app.

Adam Turaev:
Yeah. And still they're in the league on their own, right?

Jacob Eiting:
Yeah, yeah. Oh, it's staying power, the brand, their durable generational brand now.

Adam Turaev:
We're catching up, but I mean you have to -

Jacob Eiting:
You need a bird. You need some sort of unhinged avatar to...

Adam Turaev:
We'll talk about it later.

Jacob Eiting:
Big weird ads.

Adam Turaev:
We probably don't need them.

Jacob Eiting:
It worked for them! Surely that's the thing.

Adam Turaev:
Everybody has their own. But anyways, entering the market, we were thinking about, first of all, our pain point and through our own journey, moving countries, understanding what it takes to actually learn English and to break through this fluency barrier, I would say, we really targeted this specific audience. We want to address this product to people like ourselves.
At the same time, we're actually pretty data-driven when it came to choosing the market. So we looked into, I think it was English first report on English language penetration, and we saw Brazil was one of the lowest penetrated markets.
At the same time, income multiplier for these countries as English is insane. And the global mobility is also one of the key points and people are actually learning English, right?

Jacob Eiting:
Also a large country. I mean, hundreds of millions of people.

Adam Turaev:
Yeah, exactly. Exactly. Yeah. At the same time, we're actually also worrying that purchasing power is not that good. They typically don't pay for consumer subscriptions.

Jacob Eiting:
Yeah, but that's one of those things that could cause another team to overlook it, right?

Adam Turaev:
Exactly.

Jacob Eiting:
But if the fundamentals are there, then it might still be the right call.

Adam Turaev:
So we just saw all the foundation for this to work out, and we just dived into this market and just realized that most of the early adopters are most probably in this market, so we just went there.

Jacob Eiting:
Did you build an app and then only launch it in Brazil? Or did you launch everywhere and then just realized quickly Brazil's the place to focus on?

Adam Turaev:
Yeah, we launched worldwide. We didn't limit it anywhere, but just from the distribution perspective, we just focused on Brazil as the key market.

Jacob Eiting:
And did you have just English learning only?

Adam Turaev:
English only.

Jacob Eiting:
Okay.

Adam Turaev:
Yeah. Again, because we were thinking of ourselves and we were building this product pain point.

Jacob Eiting:
They say there's the famous build something people want, but even better than that is build something people need, right? And if you have huge economic upside in learning English, which you can't say the same often the other direction, right?
This is a generalization clearly, but often learning if you're an English native speaker, learning another language, it's a luxury. It's kind of a thing you do for fun, or whatever.

Adam Turaev:
This 50X cheaper than a tutor claim that everybody uses now. And actually this claim is coming from our go-to-market strategy for Brazil, specifically. Yeah, because we realized quickly, all right, if purchasing power is not that great, you have to really explain why it makes sense for them to pay $100 for yearly subscription and against a tutor that they can improve with and get a better job. We didn't sell English to them. We sold them an opportunity to get a better job.
So it was just a bit of a different angle, but all of our claims were around this, not like, "Hey, you can get better in English."

Jacob Eiting:
The apps I've seen are going to be very different to what you would see if you were in the Brazilian market, I imagine.

Adam Turaev:
Yeah, exactly.

David Barnard:
It's such a fascinating way to look at these markets as well and to look for opportunities in other markets.
For folks listening, if you're looking for a place to go, a problem to solve, look for a problem where you're selling money. I mean, essentially -

Jacob Eiting:
I really think there's only two problems to solve for consumers. It's like help them get rich, or help them spend the money they have doing something they enjoy. And maybe stay healthy, I'll say is the third one, but there's only very few things, categories, but I think that helping people make money, from my observations of RevenueCat customers, it's an easier sell, especially if the line to draw there is very direct, to we can help you get a better, more money through this career upgrade that you can have.

Adam Turaev:
What I'm saying is just a very trivial thought, sell what they need ultimately. As you're building, you always keep forgetting about that.

Jacob Eiting:
Yeah, but because there's nothing about necessarily getting a job, probably, in the product, maybe a little bit, but it's like, no, you're focused on actually language learning at that point.
When you launched, I guess obviously, when did it first hit the app store? What year?

Adam Turaev:
It was mid-2023 and it was a terrible version.

Jacob Eiting:
I was going to ask because you guys started working on it in '21, there's basically no GPT then or no LLMs then, but even by '23, we were on what, ChatGPT-4? So it's still kind of early. And then I don't know where we were in terms of voice models and all this other stuff, but it was, I think, fair to characterize it as early.
So you've probably caught a lot of tailwinds in just the models improving, but what was it like? Were you able to get the quality of product you wanted then or not? Or did you still believe anyway? And how did it go being so early on the technology curve in terms of conversational?

Adam Turaev:
Actually, even before that, we launched, I think, some web versions back in 2022 when the setup was, because of the limitations, technological limitations that we had back then, the setup was Synthesia Avatar. Synthesia was around back then. And then because those avatars are not live and NLP is not that advanced yet as LLMs, we had to build hundreds of trees manually of how conversations could go.

Jacob Eiting:
And this is before coding agents were good too, so it was an extra amount of work.

Adam Turaev:
Yeah. We had guys, I think copywriters, content creators building those trees of conversations where it can go. So it at least felt like current setup.
So an interesting fact that you see even back then, even with this setup, we really focused on making sure that the avatar is there. So we had this Synthesia avatar reacting, responding to you so that we simulate this environment to our users.

David Barnard:
Another incredible lesson there, because if you have conviction that things are going to get better, and I'll throw myself under the bus like Jacob did, early in Midjourney's days, and I remember the tweet, I'll have to pull it up and tweet out how wrong I was, but I at the time though and tweeted that, this is the kind of thing, it's going to be like self-driving where it feels like we're right on the cusp, but it's going to take another decade. Turns out it didn't take a decade. It took two years to go from Midjourney six thumbs or six fingers and stuff like that to incredibly photorealistic images.
And so if you have a conviction that the technology is going to catch up, you build what you can today with that vision of the future getting better, and it's a lot of hard work.

Adam Turaev:
You can have this conviction these days. Back then it was controversial, but now I think Anthropic, OpenAI, if you see what they're saying, their position is build on the edge of technology. And I think every other model upcoming is just proving the point that will come up.

Jacob Eiting:
Yeah, it's no longer controversial. It's not like a non-consensus bet. It had to be early though. And then I think if you look at who the winners are right now in a bunch of different areas, it's a lot of these companies that were there, not that early, but early enough and they believed and they capitalized properly and spent the right investments properly so that they could catch it and ride it. I don't want to say it's too late. There's obviously lots to do yet and stuff, but there was a big advantage to being able to sit it out and wait it out and grind it out.
For products, especially AI, and I think what we were talking a little bit off the air about the voice models themselves and how they're actually fairly good at conversing in other languages as well, which is happening to everybody.
So the core models are competing with everybody, one way or the other. But there's a second input to creating a great product, which is lessons and customer understanding and all of these things that the models only have limited access to.
But you could build that even if the technology wasn't there. It seems like in '22, '23, '24, you guys were building what do people want to use, what's our go-to-market and all that stuff. And that can't necessarily be extracted from a model, right?

Adam Turaev:
Yeah, that's true. And we still have, funny to think, we were recording all of those early customer interviews, and we still have hundreds of those interviews from 2022, people saying absolutely same things that they're saying now in interviews. So yeah.

Jacob Eiting:
I'm going way off script here, David, but I'm kind of curious, we're talking about the macros of language learning.
Another external threat/pressure is the fact that not only can you converse and learn languages with these voice models, they can actually do real-time translation. Have you found any sort of like, oh, we're just going to have the universal Babel machine now and everybody's just going to have the Star Trek universal translator, whatever?
Or has there been an acceleration or about status quo? How has the sentiment around language learning changed or not changed?

Adam Turaev:
Yeah, I don't hear this from users themselves. I don't hear this on customer interviews or anything like that, but I hear this a lot from investors, I would say.

Jacob Eiting:
Right. Who would know better? Surely the users are wrong.

Adam Turaev:
I mean, it makes sense.

Jacob Eiting:
Not the money people.

Adam Turaev:
Even us, we're thinking about it. But one of the cool stories from one of the customer interviews, an American user moved to Italy and he used our avatars to help him translate something for him to waiters. And instead of using Google Translator and their voices, he wanted his coach in Praktika to help him out. So not only help me and improve me, but also -

Jacob Eiting:
Be my guide.

Adam Turaev:
Yeah, be my guide in real life.

Jacob Eiting:
I think for a similar reason that I maybe missed how good these models were going to get, it's really easy to linearly extrapolate outcomes, but the world is very complex and it's kind of hard to tell. But I could almost imagine the desire for language learning to increase, because I don't know, hopefully AI is giving more value to in-human, in-person interactions. And you can't have learning a language, speaking as somebody who's done it a couple times, you do not get the same value out of holding an app and translating between you and the waiter or whatever. That's just a very transactional, very awkward, cold.
Having a conversation in somebody's language is a way to really connect on a human level that you can't replicate with a machine, really. Maybe if the universal translator shows up, maybe that changes. And I actually think potentially there becomes a premium on those things.
It's like more people want to travel, more people want to connect with humans in person, and this becomes a growth market. So I don't know. I'm curious what your position was. I'm not surprised that you haven't heard it from the people.

Adam Turaev:
Yeah. I mean, you describe it well. I think the easy example here is to describe that language is not transactional, and people do not learn language for transactional purposes. If you need only transactional purposes, then you can resolve it right now. We don't have to wait for any technologies.
But if you really need to belong, if you need to integrate into new culture, new society, then you have to speak that language. And even myself as a multilingual, and probably you can relate as well, you yourself even, a bit of a different person in every language.

Jacob Eiting:
Jacobo Loco, he's a real guy.

Adam Turaev:
Exactly.

Jacob Eiting:
He's totally a different person. He's more gregarious.
Before RevenueCat and I was doing language stuff for fun, I was using a service that was human tutors. And obviously it's obvious in hindsight now, that was expensive. If you want to really learn a language, you need to talk to a tutor probably every week.
And that's like, I don't know, it was like 50 bucks for an hour long tutoring session or something along that order. And that's expensive in US dollars. And you want a good tutor. You want somebody who's native in that language. You want somebody who's a good teacher and all of those things.
My belief that, going back to, not to keep knocking on Duolingo, but conversation is the constraint. In fact, that's in some ways the pre-programmed boot loader for language in humans. The books and all this other stuff is sort of things we've invented after the fact.
The human neural net is designed to intake and produce language and learn the translation later. At least that's what it feels like as a language learner. It's just the more time you spend synthesizing and processing, you just improve.
And so yeah, I mean, anyway, not to just keep selling the book, I think this is a very exciting, because I also think I've played with these tools on and off over the years, and I think it's sort of just recently that the underlying voice models have gotten good enough to really be able to do it in real time and have it feel somewhat real.

Adam Turaev:
You have to tell the difference, what's important to tell the difference from what we're doing and what you can do, let's say with ChatGPT voice or general models.
With those general models, you can role-play situations so we can have a sparring partner for your language barrier problem. But when it comes to Praktika, we're not offering just an AI companion, it's an AI coach, AI tutor, right?

Jacob Eiting:
I went through your onboarding today and you even use the, I don't know what the language scoring framework, like A1B1.

Adam Turaev:
Yeah, CFR level.

Jacob Eiting:
Yeah, which I've never used a casual language learning product that uses those. You see that if you go to talk to real tutors and stuff like this, they'll score you. But yeah.

Adam Turaev:
And we're just, I would say, simulating this environment, judgment-free environment, safe environment where you can break the language barrier. And you have to realize that for many users, again, it's not transactional purpose and it's not just a cultural curiosity, right? It's actually what limits them from hitting this next milestone, achieving this next job promotion, getting this next opportunity.
And for them, this process is really painful and you need someone around who can support you in those tough moments. You can live without learning this Spanish and if you skip this one-week lesson, you're fine. But for those guys, this is probably the only opportunity they actually get to learn this language and to get this opportunity in life. So this is something what I think what we realized early on and what allowed us to really address this pain point, because we live through the same pain point.
And for us English as three co-founders, Anton and Ilya, we were living through this and we actually broke this language barrier and we realized we're having access to this next level opportunities. We're not local anymore, right?

Jacob Eiting:
Right. It's basically the global, from my perspective at least, the global language of capital is English. The global language of technology is English. Even firms that work in a locale, often they still work in English, even if it's 80% revenue cap, I think is probably majority English as a second language.
Yeah, it's got to be at this point, but still English is the common language. And it seems like, especially given that the labs are both American companies, primarily in English, and much of the research is, with the exception of obviously there's a lot of research going on in China too, but seems like the lingua, the common tongue is going to continue to be English into the future.
It does not seem like a trend that's going to reverse our plan.

David Barnard:
I'll bring us back around here soon.

Jacob Eiting:
Yeah. We should talk about app growth and stuff, but the language learning market's really interesting.

David Barnard:
Let me interject before you ask that next question. I think a lot of folks listening will be like, "Oh, they're talking about language learning. A lot of that stuff is really specific," but I think there's a lot of things in there that you were talking about that people should go back and re-listen to the last 10 or 15 minutes, but then think through how each of those things we discuss and each of those things Adam pointed out come back to the jobs to be done of this app.
You're trying to get somebody who doesn't speak English a better job by speaking English. And so when that's your goal, not Jacob's learning for fun and those kind of things, when that's what you're selling as an app, you make very different choices. And that applies to any app you're building. What are you really trying to solve for your customers and stay focused on that, even though learning for fun and there's all these other things that can pull your attention away and take the product a different direction.
But that whole conversation was really centered around that this is the core job to be done and that's why you stayed so focused on it.

Adam Turaev:
I mean, I think it's also very actual for this AI era specifically when everybody is very hesitant about what modes are, et cetera. I think it's still in owning the outcomes and being really focused on those outcomes because if we're talking about achieving those outcomes and serving those outcomes to users, then things like context, harness, opinionated user experiences, all of those things just become prioritized. And I think that's what makes a product different these days, just simple product taste, I would say.

Jacob Eiting:
Yeah. I mean, I think this is a big debate happening all over right now is how much is there left at the boundaries between the models and the consumer? I don't know, it's kind of self-serving, but I do believe that there's sort of going to be the models of the ChatGPT app, Claude app, whatever, the generic agent platforms are never going to be able to fit all of the things into them. There's just not enough surface area. I think mentally it's too complex.
Sure, the agent could probably synthesize something of moderate quality, but your team will always be able to take whatever the agent can do and then add to it and make it better, more efficient, more targeted, more communicated clearly, I believe at least, maybe we're heading to a true singularity. But I was actually going to ask you, this technology in your product is part of somewhat displacing the existing infrastructure of private tutoring and things like that. How have you thought about schooling?
Because I imagine, especially in these countries where English has an economic benefit, teaching it in school seems to be prioritized because I would even think... I had language teachers at my school and your app would outperform them today. You know what I mean? It would be better if they just locked everybody in a room and had them do Praktika, than if they actually take English or actually take a foreign language.
But I'm curious, how do you think about the education system as part of your ecosystem?

Adam Turaev:
Personally, I would love to be part of it, but I think they're really lagging behind because of all of this regulation.

Jacob Eiting:
Well, obviously the teachers' unions aren't going to necessarily jump at the excitement of being replaced by an avatar.

Adam Turaev:
Yeah, exactly. But I think what we need to realize that it's not going to completely replicate or substitute tutors, et cetera, because that's this unique knowledge, unique methodologies. I truly believe their mission is even more important right now because we have to work in partnership, I would say, to build AI together that can scale their knowledge, scale their methodologies.

Jacob Eiting:
Yeah. I mean, I'm sure every tutor would love to spend 10 hours a week or 20 hours a week there.

Adam Turaev:
Exactly.

Jacob Eiting:
If they can't do it, their clients can't afford to pay for it and they can't afford to do it. They don't have enough hours.

Adam Turaev:
Exactly. And what they can do now with AI, if used right, if not feared of, but used right, they can scale it. They can serve not 20 students, but they can serve 1,000 students or more and do it one-to-one face-to-face.
We're getting to the point where you can clone your voice, face, build your own avatar, feed it with your context, your special source to pedagogy, and it will serve your students in your own way. And it's unique. That's what makes it special.
None of best models out there can do this because you have your own special DNA to what you're doing, right?

Jacob Eiting:
It seems like, and David, I'm going to jump back onto the schedule here because you had a question about the avatars. So I think that's something that's somewhat opinionated about what you guys have built is that you do have these Pixar-esque, at least seems like the most recent ones are Pixar-esque.
And I'll say from my experience, it feels a little uncanny at first, but in terms of, it certainly makes the content marketing, it makes it much more visually understandable to see this thing talking. And also, I think we're just getting out of the uncanny valley for live avatars. I've just this year started to see the first ones that aren't very cringey.

Adam Turaev:
Oh, I think still there is to some extent -

Jacob Eiting:
We're not there yet.

Adam Turaev:
Creepy, right?

Jacob Eiting:
I can see we're not far. We're not far. What was the rationale behind going hard on that and making that such a central part of the experience?

Adam Turaev:
It might be controversial, but it was never with marketing or distribution in mind. It was, again, our understanding that language barrier to some extent is not only about the skill issue. A big part of it is a psychological issue of you talking to another human in a foreign language and you have to think fast and it's super stressful. And I remember on one of our onboardings, the key insight was for us as one of the reasons why they're here was fear.
One of the most clicked button was fear is the reason why I want avatars to teach me language, or I want to rehearse.
So that's why we built them to simulate this environment of you talking to another human. We knew that it didn't have to be super, super cool technology that simulates human to 100%. It just needs this feeling that, all right, somebody's watching at me right now, and it gives me this simulated stressful environment of me talking to another human. And I rehearse this feeling, right?
So that's why when people work with Praktika and then they have to speak this language in other cases, but with real human, it's much easier for them. So that was the key points. And again, I remember when we were raising our first round, that was I think one of the most questioned parts of our product. But now I think it's pretty obvious why you need it.

Jacob Eiting:
Yeah. And I don't know what your production process is like and what it takes to launch a new avatar, but I have to imagine in one, two, three years, the price of producing an avatar is going to drop, the quality's going to go up, you'll be able to just do more and more with this.
So it's probably another example of... I know, David, I've seen some avatar launches recently where I'm still like, this is kind of rolling my eyes a little bit. But again, I need to check myself because while I'm rolling my eyes and cringing, if I just wait 12 months, it's going to be there.

Adam Turaev:
I can easily imagine. And we're not religiously into those Pixar-style avatars, but as of now, this is the best way to give this experience to users without creeping them out.

Jacob Eiting:
Right, because, if you try to be too human, you really get into the uncanny valley. So it's like the cartoons almost make it easy. And it's probably why if you see AI-generated dramas and stuff like that, a lot of times they use a cartoon Pixar style because it like-

Adam Turaev:
That's why I also see Muse, Dots and other guys not launching human avatars.

Jacob Eiting:
There were some really cringey attempts at human avatars 30 years ago and I think people are still burned from those.

Adam Turaev:
Yeah, yeah, that's true.

David Barnard:
Well, I did want to dig into growth because I think a lot of people will be fascinated by how quickly and how you scaled up. So I know micro influencers were a big part of that scaling. So I'd love to hear how you got started with that and then what worked.

Adam Turaev:
When we started actually, we didn't think, all right, we need micro influencers. It was more like my formula to it was I don't have budgets for large influencers in the region. I need someone like a rising star. So my approach to it was I went into a platform called Modash. It was early days back then as well. And what it allowed me to do, it allowed me to limit the following because that was the main indicator back then. If you have big following, then you're charging a lot.
So I limited this to less than 5,000 followers. But at the same time, I filtered out those who are getting a lot of use these days. So the algorithm is prioritizing those authors. And that's how I found one of the influencers who had been back then hitting tens of millions of views on their educational videos. And the guy didn't have any email for collaboration, nothing mentioned, and he didn't even realize what's his [inaudible 00:35:57].
So yeah, I just reached out to him right in TikTok and just told him, "Hey, we're building this app and it seems like we have a good match, so do you want to work together." He's one of our employees now. He's within our team, Gustavo, absolutely super, super talented guy and still building videos for us.
And he become to some extent a face of Praktika back then, still running our social media.

Jacob Eiting:
You said you searched for low followers and low distribution.

Adam Turaev:
Low followers, but high reach.

Jacob Eiting:
High distribution. Okay. So it's probably a pretty good indicator that you're catching somebody on the up, right? They haven't fully acquired a bunch of followers yet, but some of their stuff's gotten picked up by the algo.

Adam Turaev:
And there are still so many influencers. It still works, I suppose. It can be used as a framework, not just targeting micro influencers, but filtering those out who are hitting a lot of views with low following to something.

Jacob Eiting:
It's the content. It's the content that's driving the views, not the existing follower base, which is probably more valuable. Especially when you're trying to price this asset, you always want to find assets that are under-priced. So if you can find somebody who hasn't really discovered how good they are yet.

Adam Turaev:
It's not even about under-priced, but it's more that you're approaching someone on their pick even before they're picking.

Jacob Eiting:
Right, I didn't mean literally you have to pay them less. Maybe you do, maybe not, but it's like buying stocks or whatever. You want to get in early and that's when everybody wins, right?

Adam Turaev:
Yeah. It allows you to be early. Yeah, exactly. Yeah.

Jacob Eiting:
Did you do that across a bunch of people? Did you do a whole bunch and then had a few winners or was it -

Adam Turaev:
I had a long list. Back then you didn't have all these automations you have these days. So as founder, I think I had spent probably a month doing this. It was a full-time job. I just kept sending those emails, messages to all these influencers, and a few of them just worked out. And why I mentioned Gustavo, because specifically with Gustavo, he built one of the videos, he sent it over to me and I was like, "Eh, I don't like this video, but do whatever, you can post it."
And he posted it and this video just blew away 10 million views in Instagram, another 10 million in TikTok, another 10 million in YouTube. That was the moment when we just scaled really quickly from one million ARR to 10, 12 million ARR.

David Barnard:
I wanted to dig into that specific story because that is insane. So you went from one million in ARR to 12 million in ARR in a single month. That was the Gustavo video, and then you used that same format over and over again? Or how did you do that in a single month?

Adam Turaev:
The video that we made back then is now the main structure of the video for language learning apps these days. So yeah, he went for, again, because he's Brazilian, he knows what's the pain point actually. He knows it deeply. So that's why he built a video where a coach is just correcting mistakes. You just have little conversation.

Jacob Eiting:
Oh, I know this format. Yeah, yeah, yeah, yeah. Where somebody says it and then the coach is like, "No, that's a duh."
Wow, why is that so addictive? I know exactly. It's a good hook and I'm not even sure why. It doesn't seem that... I can see why you went "Eh."

Adam Turaev:
Yeah. And for me it was, "Hey, why don't you speak about the app? Probably you want to showcase the app itself or anything like that, but he went, "Believe me, trust me."

Jacob Eiting:
Yeah, yeah, yeah.

Adam Turaev:
I was like, "All right."

Jacob Eiting:
You see some viral videos around this sometime where it's like, I think there's something in humans. When we see somebody doing something wrong, we have this sort of impulse to correct that I think is very...
There's that famous meme video of the person putting the shapes in the shaped holes and it's like, "No, you put it in the round hole."
You know what video I'm talking about?

Adam Turaev:
Yeah, yeah, yeah.

Jacob Eiting:
And it makes you go crazy and it's very hooky because you're screaming at the television. I think the videos you're talking about, I think it's kind of a similar thing is where you hear somebody make an obvious mistake and you're probably targeting, you're not targeting the basic, basic beginners, necessarily. So you want people who have... They've taken a few lessons, so they're like, "No, that's wrong and I'm smart," which is awesome. That's very interesting. That's very interesting.
It's like everything's a big exploration of a state space. So there's all these variables, whatever. And then somebody just tries a thing that just falls into a new global optimum like that. And then yeah, I'm sure all your competitors are using that format now as well.

Adam Turaev:
Yeah, they do.

Jacob Eiting:
Well, you explained it while you could, right? You get the alpha out before it's shot.

David Barnard:
Back to the mechanics of that though, so Gustavo did 30 million views with a single video across the three different platforms, but I imagine that alone didn't drive you from one million to 12 million in a single month.
How many creators were able to replicate it? Did other videos do 30 million views? How did that month go going from one million? And it must have been really fun internally. I mean, one million to 12 million ARR in a single month is just insane.

Adam Turaev:
I would say we didn't realize what was happening. We were just, "Hey, seems like it was working. Let's just scale it as fast as we can."
So we started just replicating those videos with other creators in other countries as well that would allowed us to scale that process.

Jacob Eiting:
Can we do that in every sort of source language, target language? There's a million and every different avatars, whatever, different people. So very scalable.

Adam Turaev:
And just historically, that's why we see a lot of launches these days where people are very targeted on specific countries, but that specific thing, we saw the opportunity to scale this format relatively quickly across the globe.
So we started just replicating with various angles and stuff.

David Barnard:
And did you put any money behind that? Did you run it as an ad or create ads in that format? Was Bend part of the equation that month that really blew up?

Adam Turaev:
We put some budget behind it, but it was a tiny budget, I would say. First days we were crazy, I would say, with literally close to zero-cost CPMs, CPIs and cost per trials and purchases and stuff.

David Barnard:
Was Gustavo an employee at that point or what was your arrangement with him? Because I know a lot of influencer deals these days, or micro influencer even, it's like you'll give bonuses for a certain number of views. So did Gustavo really cash in? Was he an employee or -

Adam Turaev:
I hired him immediately like, "I need you, guy."

Jacob Eiting:
Put a ring on it, as they say. Get over here. That's amazing. It's amazing. I just love that you didn't like it too.
I always find it's so hard, especially I think as the... I don't know, this happens a lot at RevenueCat. There's a lot of stuff that I would've killed that we've done that I'm glad we didn't, and some stuff that I didn't kill that I wish I had.

David Barnard:
You got to take the twings, Jacob.

Jacob Eiting:
Exactly, exactly. It's like you kind of have to explore a large space to see what's going to work. And if you shut down everything that you don't logically think is going to work, you're going to miss these opportunities. And so yeah, there's this balance between, okay, what kind of risks do we take as a brand versus what, do we editorialize too much?
But often I also feel that the "conversation," the meme mystery, everything moves so fast now that you can try stuff and if it doesn't work, even Revenuecat, we can try a thing. If we do some crazy idea from our marketing team and it doesn't go anywhere, we just pretend that never happened. We'll just move on. It's not a big deal.

David Barnard:
From our marketing team, you mean Rick?

Jacob Eiting:
Some individuals. Some individuals produce things more out of distribution than others, but it's a good lesson though. I mean, it's a lesson I learned too early, which is you got to trust your gut obviously, but then as a founder, if you force everything to go through your approval filter, you're going to miss things that are just your blind spots.

Adam Turaev:
No, I mean that was always my rule since the beginning. Do not tell influencers how to do their job. So you are hiring them to make -

Jacob Eiting:
I was going to say, if you were that influential, you would just do the videos yourself.

Adam Turaev:
Exactly. Yeah. That was exactly my way of thinking. So if you're hiring them because they're getting an amazing reach, so let them do their job. So that was the point.

David Barnard:
What does your influencer program look like today? I know you have some folks in-house like Gustavo, but you're still partnering with a lot of influencers. What does that look like today? I know things have changed a lot in the past couple years.

Adam Turaev:
I mean, the approach is same. I really want to have them in-house. I always prefer that. And we have this creator program with influencer lead, influencer coaches running various countries. I was always opinionated about that.
Those are X influencers that were building videos for us, and now they're running those programs because they know what good content is, and that's the key skill you need to have because you can sort the management stuff out and it's not that complicated, especially now. So my influencer lead is someone who we'll work with in Poland, and she was just making videos for us, and she had done an amazing job, and then we hired her as an influencer lead, and she's now running the world programs. She's doing an amazing job.
We have this influencer lead, and then influencer coaches running various countries, and each of them just run a roster of 20, 30 influencers, and they just rotate various angles.

Jacob Eiting:
When you say it's in-house, these leads are in-house, but the influencers you're working with are still contractors?

Adam Turaev:
Yeah, leads and coaches are in-house, but influencers.

Jacob Eiting:
There's an aspect of it that's in-house, but then you're still farming out. Going back to the point of let artists be artists, you have to go out. You don't want to have them on employment. You don't want to have them be their boss necessarily.

Adam Turaev:
But I'm always looking around for talented guys.

Jacob Eiting:
Right. To build your bench, build your in-house. It's interesting.

David Barnard:
How many actual influencers do you have in-house now? Because you have these coaches and people running the program, but you have Gustavo and others who are actual influencers in-house, right?

Adam Turaev:
Yeah. This is a separate stream. Gustavo is now running as internal influencer, and we now have this creator program with, I think, more than 100 influencers in the program right now, and we just keep scaling it, keep rotating, because there are so many opportunities within various countries. Even though we have 100 of them already way to scale still.

Jacob Eiting:
This is a really naive question, I think, but is there a platform that you manage this stuff with? Is there a-

Adam Turaev:
Oh yeah, sure. I think everybody's now doing this through SideShift, and we use SideShift as well. We just build a few our own internal add-ons to run this, to do more comprehensive analytics around it, and some reporting.

Jacob Eiting:
Your specific goals with the program.

Adam Turaev:
Yeah, exactly.

Jacob Eiting:
But there is a platform, especially because it's marketplaces, right? It's always nice when there's a centralized tool for these two-sided things.

Adam Turaev:
Yeah, SideShift is doing an amazing job there.

David Barnard:
I'm curious because I feel like every single person I talk to has a different philosophy and a different structure. How do you typically structure the deals? Are you paying per view? Are you paying a flat fee per video with bonuses? Are you paying per affiliate type deal? And maybe it varies across, but I'd love to dig into some of the ways you structure that.

Adam Turaev:
Because we have those influencer coaches structure, it allows us to be flexible. So we're just optimizing for the best talent, and it sometimes requires some additional structure to the deal. So it's not that, "Hey, we'll only do fixed fee plus flexible bonus." It really depends. If there is a good talent, just do whatever to hire this person. So that's the approach.
But typically, yeah, nothing too special, like fixed part, and then there's a bonus for number of views they're getting.

David Barnard:
Yeah. How many of those that go viral do you end up either using directly as ads or turning into an ad format that you generate? And then I'm curious, the paid ad side of things and how you synergize... Synergize.

Jacob Eiting:
No, you got to leave it in, editors. Don't let him, don't let him...

David Barnard:
And how you work those two programs together simultaneously?

Adam Turaev:
This is paid ads and organic content, this creator program, but those two are separate streams. We don't mix them up because again, this allows us to be true to our principles of letting influencers we work with to come up with ideas on their own, to come up with those angles.
Because I think at some points there's a fatigue internally around what can we offer, what angles can we do? And every new influencer upcoming is a potential for this new way of thinking. So we'll let them do the job and we do not mix them up with paid acquisition. But if there's a cool video that's hit virality, we then share this with paid team, and then they reproduce this video with their own roster of internal influencers or contractors who work with content creators, et cetera.

David Barnard:
Yeah, that's fascinating. Before we move on from the viral and UGC and all that, I know you've had a few kind of misfires along the way with unhinged ads and stuff.

Jacob Eiting:
I was suggesting doing unhinged avatars before, and you got real quiet. Now I know why.

Adam Turaev:
There was a point where those unhinged avatars were going viral in upper funnel, but then there was a very reasonable thing like, "All right, let's add them in the app." But it didn't work out well. It actually caused more people not being happy about this, and not understanding why my avatar is cursing. The image is fine. I want to talk to this avatar, but why is it cursing?
That's why we just decided for us, we don't want cursing avatars. And it comes down to this principle that we're not serving AI companions, we're serving AI coaches, and you want your coach to be a little bit more probably serious.

Jacob Eiting:
Also, just a lot of people probably just don't get the joke. You know what I mean? A lot of people aren't on TikTok and stuff like this, so they're just like, "Wait, what's going on?"
Going back to our favorite green owl, but the character of, what is Duo, Duo the Owl, is much more crazy and unhinged on their social. But if you go in the app, I don't think it has that level of insanity. But it's a good example of... We run into this often, RevenueCat's got this quirky, weird cats, and then people are just throwing out ideas, and then eventually something leaks out and it's like, "Okay, there was an invisible line and we need to come back," and I have to once in a while be the party pooper that's like, okay.
We struggle because I think there's so much virality and enjoyment and stuff in the quirky, the fun, the unhinged, but it does have costs if you don't keep it in the right domains and in the right context. It's what makes humor difficult. It's like a joke can land very differently depending on the context, and it's really difficult to find those lines.

Adam Turaev:
I think at the end of the day, it's absolutely fine to draw the line between, "Hey, here's my product guardrail, but for growth, I can have some separate-"

Jacob Eiting:
Yeah, that's maybe a nice clean boundary for it. And same with us. It's stuff that's on Twitter or whatever. And again, it goes away. It's ephemeral. It's things that are ephemeral is one thing, but when it's permanent, my favorite's fine, if we're good, we're getting away, but we have these RC friends that are awesome, these little characters. But at some point we decided they needed to be in every page in the dashboard.
It wasn't decided, it just happened because we had them. And then one day I woke up and it looked like a McDonald's inside the product. I was like, there's just characters everywhere.
I was like, "Hey, we got to dial this back a little bit. It needs to have a little light taste."
But it's hard. It's hard to set that line. But you call it guardrails, which is what you would call something you gave an agent, but it's almost the right thing. It's what are the guardrails for these different domains, needs to be defined by somebody.

Adam Turaev:
Yeah, exactly. Exactly.

David Barnard:
I'm curious how you specifically, because we've talked about finding those outlier crazy things. How do you balance that? And is it just like we let social and growth be totally unhinged, or is there some cap on that, and then the product has a cap, but it's just kind of a lower ceiling of how unhinged we can be inside the product?

Adam Turaev:
When it comes to product, there are very clear boundaries. We just agreed internally, that's what we're doing, that's what we're not doing. And it's really just a one sheet of description. It doesn't have to be complicated.
When it comes to growth part, we'll let them do the job. We let them create. We don't limit them, to be honest.
There are some basic limitations, but the difference is pretty dramatic, I would say. In growth, being true to our principles, we let influencers to create and to come up with pretty much any angle that they want to.

David Barnard:
Before we wrap up, I do want to dig into a little bit of where you're at today. We talked about scaling from one million to 12 million ARR. What's happened since then and what does the team look like?

Adam Turaev:
We are now at 50 million ARR, reaching almost half a million paying subscribers.

David Barnard:
Wow.

Adam Turaev:
And still really in team, a little bit more than 40 people. So our AI error revenue per employee is more than one million. So yeah, scaling right now, actively focusing on, I would say, making those coaches probably a little bit more proactive so that the idea is that we want them to guide you because it's a painful process. So being inspired by what Muse is doing, Dots, right?

Jacob Eiting:
Yeah. It feels like harness engineering is a big thing, basically what you seem to be doing.

Adam Turaev:
Exactly. But how we are different and how we position ourselves is like those are accomplisher AI. They do tasks for you, but what we do, we guide you and we improve you as a human, because the quality of output is really depending on quality of input. So we are working on the side where we work on quality of input to make sure that humans using those tools are using them right basically, right?
And talking about security of those AI models, I really do think if you want AI to be secure, safe, we should really care about who are using them and how they're educated, what's their approach, what's their angle of using AI. So that's how we see our mission in this ecosystem to make sure that we're building coaches that are guiding you to improve as a human. So we're starting with language learning, but then scaling to other subjects as well.

Jacob Eiting:
It's a great example of just what an agent, an agentic version of a traditional category looks like. And it's kind of interesting because I think you started with a very sort of direct inroad, which is have a conversation, say a thing, get a reaction, but the problem quickly becomes much bigger, which then encompasses all of the other things in apps that we've been working on for decades, which is how streaks, retention, when somebody comes back.
I'm guessing you're thinking about just handing more of that over to the agent and letting it control more of that experience?

Adam Turaev:
We are already doing this. When we want to set a daily reminder, we're not giving an interface to set this up. Yeah, we make tutor ask, "Hey, do you want to set a reminder? I can set this for you."
So that's how we see the future, right?

Jacob Eiting:
Interesting. It's going to have all kinds of interesting implications for just measuring things and testing your product and things like that because every agent's acting a little differently and things like this.

David Barnard:
The last thing I did want to ask about was, and I feel like this is such a big topic these days, with zero marginal costs going away, how do you think about balancing the cost of generating these avatars, of having live interactive conversation and all the inference costs against what you charge and freemium and -

Jacob Eiting:
Quality too. It seems like inference costs and quality are sort of inversely correlated.

Adam Turaev:
Look, I've been always focused on delivering the best quality to our users, and I think it always pays back in the long term.
So we just have to prioritize making sure that the quality is there, because as you see all those models, they're getting cheaper over time. So just prioritize making sure that you're delivering the best quality of your product. I'm pretty sure models will get cheaper and it will allow you to balance your economics. Even now with this latest Luna launches, I remember we were looking at 5.6 Luna, which was 50% cheaper. We were like, "Yay, our unit economics is better." And we're profitable, by the way.
You can be at this scale, be profitable and still scale, so you don't have to burn money. So I'm not saying just keep burning, raising money and then focus on quality. You can focus on quality and be efficient.
So then this Luna was launched 50% cheaper, and then I think two or three weeks later, Luna 6, and it's 50% cheaper than this 50% cheaper model. So it's getting cheaper that fast.

Jacob Eiting:
Yeah, it's happening.

Adam Turaev:
Yeah.

Jacob Eiting:
And it seems like I don't see a limit to it. We're getting the data center build out. At some point, you have as smart of a tutor as is useful. Do you know what I mean? I don't need a super intelligent Spanish tutor. I need a pretty intelligent Spanish tutor. I don't need one that can solve Navier-Stokes or problems and things like this, right?

Adam Turaev:
At some point it will help me build a rocket, right?

Jacob Eiting:
Yeah, exactly. So at some point, you maybe don't need to ride the intelligence curve anymore. You can say, "This is good enough," and then that price should just drop.

Adam Turaev:
My bet is that end game is those models just being capable of running on device, like LLMs, full stack, being able to run on device. And that's where we just get back to the point where -

Jacob Eiting:
You can distribute the power, the cost to the device, or it just becomes as cheap as databases are. Nobody really thinks about how much... RevenueCat has really big databases and they're expensive, but they're not like AI is expensive. It's because databases have been optimized for 40 years to get cheaper and cheaper. And I think the same thing will happen to AI. It's just going to get cheaper and cheaper. We're going to drop price and it'll drop to the cost of a kilowatt to power the AI. You know what I mean? That's going to be the marginal cost at scale.

David Barnard:
Cool. As we wrap up, I did want to ask you three questions I ask every guest now.
What was the most impactful experiment or change you made in the past year, your biggest win?

Adam Turaev:
Surprisingly, our biggest win was moving from cross-platform Flutter to iOS native was our biggest win.

Jacob Eiting:
Why was it such a win?

Adam Turaev:
We didn't expect it to be such a big winner, to be honest. We were just thinking, we were just praying out there, let's not lose our metrics that we worked on those years. But I would say stabilize the app to provide users with the best experience possible for their device, and it just paid back in improving metrics. And it really, surprisingly, it improved across the board, all the conversion metrics, engagement metrics.

Jacob Eiting:
The app just felt tighter.

Adam Turaev:
Yeah.

Jacob Eiting:
How did it affect your development process? You go faster now that you're native?

Adam Turaev:
I would say same. We're pretty fast, so we run a lot of experiments, and I wouldn't say it changed a lot.

Jacob Eiting:
I think it's the sell for Flutter, React Native and all these. In the early innings, I think of the AI revolution, it seemed like, "Oh wow, these technologies are really super important now because the agents are better at them."
But then we kind of passed it very quickly where it's like, well, now you can just have the agent write Native iOS and Native Android and it's almost an implementation detail at this point, which is kind of wild. And then of course, I think Flutter also had some design choices that made it less fit for this new world than even React Native.

Adam Turaev:
I mean, it still does the job, right? So you can kill two birds with one stone still, but if I was doing this over again, I would just go with iOS Native first, and then we'll just build Android player in that -

Jacob Eiting:
Cloud Max Pro subscription and two apps and just let it rip.

David Barnard:
I think people are still making those kind of decisions. Do I jump in Native?

Jacob Eiting:
Just give a Codex agent both apps and just be like, "Convert this, and make no mistakes."

David Barnard:
It's that easy, Jacob.

Adam Turaev:
Make the best app.

Jacob Eiting:
For sure.

David Barnard:
So what's the worst experiment, the biggest fail of the last year?

Adam Turaev:
I mean, I think we took it through with those unhinged avatars. I think that was the biggest fail, to be honest.

Jacob Eiting:
Permanent brand damage.

Adam Turaev:
Yeah, yeah.

David Barnard:
So you implemented unhinged avatars in the app and it just freaked people out and they just had a bad experience.

Adam Turaev:
Yeah, but they didn't like it. They didn't like it at all. And probably it's just our thing, because our picture on the market is probably a bit more good guys, I would say. So probably those unhinged avatars just didn't land.

David Barnard:
All right, last question. Growth would be easier if...

Adam Turaev:
There are no copycats.

Jacob Eiting:
Everybody's a wannabe monopolist. Competition is for losers.

David Barnard:
Yeah. That's awesome. Well, man, this is so much fun. As we wrap, is there anything you wanted to share? Any jobs you wanted to shout out at Praktika?

Adam Turaev:
We are hiring a web engineer right now. We are hiring a user acquisition manager as well. So yeah, we are welcoming mad people who are ready to work 24-7 to win this market. So if you're that person, please welcome to the team.
But yeah, I really enjoyed this conversation. That was a lot of fun.

David Barnard:
Yeah, likewise. Thanks so much for joining us.

Jacob Eiting:
Thanks, Adam.

Adam Turaev:
Thank you. Thank you.

David Barnard:
Thanks so much for listening. If you have a minute, please leave a review in your favorite podcast player. You can also stop by chat.subclub.com to join our private community.