Three months ago I did an interview for the Hustlers’ Podcast. After watching it, some of my relatives told me they finally understand what I actually do for a living. That felt like a win worth sharing.
So here’s the full transcript. We talk about scaling Aithor from zero to $1.5 million MRR, how influencer marketing drove most of that growth, and my recent dive into building apps with vibecoding tools.
Originally published on Hustlers’ YouTube Channel
Introduction
Host: Anatoly, hello. Thanks for coming. Good to see you. It’s really interesting to talk with you. I know you from the side of how you’ve scaled products on the international market. There’s the Aithor case, which is interesting to learn more about. And in general, what have you been doing before, what are you doing now? As far as I know, you’re currently traveling in the US, meeting people in the Valley, which is also interesting. Could you introduce yourself? What did you do, what are you doing now?
Anatoly: Anatoly Terentyev - marketer, startup guy, adventurer, rebel, stickler for details, genius. I’ve spent my entire conscious career in marketing, promoting very different directions. At some point I understood I needed to transition to working purely with digital products. I worked at different companies. Mostly they were all startup-like attempts to launch something new. I don’t have much corporate experience. Well, there’s Skyeng if we’re talking about something big. Though at Skyeng I was also handling new directions and launches.
But if we’re talking about the main cases, definitely Aithor is the biggest case I’ve pushed from zero. We made 15 million in revenue, 15 million dollars in revenue over 2 years. And it all started from nothing. I remember the moment when we had like 200 dollars per week, and then it became 400,000 dollars per week. Those were the numbers. I managed to build all this, and it’s my biggest case.
This all happened pretty fast. And it’s important to say that by Valley standards it was practically without investment money. Well, relative to the Valley, because in the Valley a million dollars is very little. I’ve been traveling here, meeting founders and investors, and I see the difference. For scaling to make 15 million in revenue, I think some projects could easily spend two or three times that. That’s pretty common in the Valley from what I’ve seen.
My Marketing Background
Anatoly: Actually, I love all marketing. I do it all, and find everything super interesting. I started once with paid ads in Russia, with Yandex, then Google, then Facebook, VKontakte back then. Then I started getting into SEM. There was a lot. I’d say I’m a marketing polyglot - I understand everything overall and did it all.
Lately I’ve been doing a lot of influencer marketing, because that’s how Aithor made 15 million - the biggest months were one and a half million per month, and it was influencer marketing, TikTok, Instagram. In total we got over 1 billion impressions. Then we just stopped counting, reached a billion and just… that was it.
And this is actually the most interesting experience, I’d say. Why? Because CPM will be expensive and always will be. Content has difficulties. And TikTok itself is actively banning for content now, tightening the screws. Reddit, which was a great platform for traffic, is becoming very difficult. From recent experience, my team and I were registering accounts on Reddit, and everything gets banned. You just create an account and get banned for a message like “What beautiful code.” And that’s it.
Why? Because people now use Reddit less for traffic pumping. Reddit doesn’t like this, TikTok doesn’t either. And I think they’ll tighten the screws a bit more. But influencer marketing can’t be stopped because they can’t ban influencers who create content and essentially run the whole platform. This is the expertise I’ve developed over the last couple of years.
The Agency Days
Host: I saw on LinkedIn your first job was as founder of an agency.
Anatoly: Yeah, I had my own agency. I sold setup for Yandex Direct and landing page creation. That was a super popular offering. I sold it for an average price, like 70 to 100,000 rubles (~$1,000-1,400 at the time) - that’s what landing pages plus Yandex Direct cost.
Then Tilda appeared. Tilda democratized everything. And when landing page prices dropped significantly, well, Marquiz appeared and finished off that whole direction. It became completely impossible to sell this for any meaningful money to the average market in Russia.
So either you had very expensive clients willing to pay a lot, but it’s not always easy to generate them, or just people wanting you to do it for 15,000 rubles (~$200). And you can’t really work with them. So these guys went to Tilda. They went to Tilda, they went to Marquiz, and services popped up around that.
The agency business is not the simplest business, actually. But it teaches you a lot. First, because you have a constant flow of clients coming and going. Average client lifetime in agencies, I think for everyone it’s about 6 months.
Host: So how long does a client stay with you?
Anatoly: 6 months, yeah. Why? Because there are two scenarios. Either everything goes really well for them and they don’t need you anymore. That happens, especially if it’s a physical business - they have capacity limits. Or everything goes badly for different reasons. Sometimes you as an agency didn’t do enough, or their business is struggling. Bad offer, who knows.
I’d say the agencies that do well are those who count very strictly - they know their hourly rate, they know what it should charge for. Then it’s a normal business. The problem with small agencies is they don’t fully understand the cost of an hour and don’t know what to charge for. So there’s a mismatch - they sold hours very cheap and then had tons of work. That’s a constant problem with small agencies in the market. Always happens, always will.
Joining Aithor
Host: So with Aithor, what exactly was the stage when you first joined?
Anatoly: I was basically the first non-developer hire. Dima Afonov had started everything, and he had one backend developer already helping him with some code. When I came in, there was literally nothing else. For a long time, we didn’t even have a product team. Dima and I would just call each other and figure out the best approach on the fly.
Host: What was the product like initially?
Anatoly: We had the first version of the website - it was called SA Pro back then. Written as S-A-I, then “Pro”. Good domain, actually. But we had to drop it because it was too close to another domain. The thing is, if someone listening doesn’t know what the product is, it was written through AI - undetectable. That’s the whole point, right?
So here’s the interesting part: when I arrived, we didn’t have this positioning yet. We didn’t call it “undetectable writer.” I found that angle.
Host: How did that happen?
Anatoly: I set up Google Search Console and we had barely any traffic. But in that little traffic we were getting, I noticed the site was ranking for the search query “undetectable writer.” Someone was actually searching for that. And I thought, “Wow, this is really concise and powerful - we should use this.”
The original offer was just “write faster,” right? Which, honestly, a lot of AI projects do. “Do it faster.” And sure, that works to some degree, but people don’t always need things faster. Sometimes they need better quality, more creativity - lots of things. That positioning just felt weak to me.
The Influencer Marketing Playbook
Host: How did you get into influencer marketing?
Anatoly: Most of the things I discover come from Twitter, honestly. I always tell people to read Twitter. It’s literally made me earn way more and discover things I never would’ve known about.
I was scrolling Twitter one day and I saw a post about this video - “Secret Apps, Number 44” - and the post was like “Look, they do this, and you can find them with this exact Google search.” So I looked, and the video had massive reach - I checked and got like a 1.5% conversion rate on a video with 500,000 views. I thought, “Wait, they’re getting traffic from TikTok from this?” That’s interesting.
Also, around the same time, I looked at this app called Gauthmath - it’s affiliated with TikTok directly, promoted by ByteDance themselves. And I thought, “How does TikTok promote itself on TikTok?” I looked, and they do it with influencers. No links, no “click here,” just mentions. And because there are so many videos, each one pulls in huge numbers.
Host: And you tested this approach at Aithor?
Anatoly: Yeah. I had my first attempts before. We found a Russian teacher and paid her like 1,700 rubles (~$25) for a TikTok video. When I heard the price, I said, “Obviously she quoted us too low - we were probably her first customers.” This was winter 2022 (early February), right before TikTok got shut down in Russia.
And the first video did a million-and-a-half views for 1,700 rubles (~$25). The traffic kept growing, and when we asked people where they heard about us, they said TikTok. Unfortunately, we couldn’t scale it because TikTok got banned.
But I had complete confidence in the approach. When I interviewed at Aithor - Misha, Herman, and Dima were asking what we should do - I said straight up, “We need to make TikToks exactly like this. It’ll work.”
Host: And it did work out?
Anatoly: Yeah. I have a case where we paid 400 dollars and got 9 million views. Those kinds of returns. At one point we were putting out roughly 400 videos a month. A lot of them hit. The best part? Friends would text me like, “Oh, is this your project? I just saw this organically.” That organic reach, people discovering it without a direct link - that was the most satisfying part.
Tracking Influencer ROI
Host: How do you track attribution without links or promo codes?
Anatoly: The way I see it, it’s like carpet bombing. You’re doing massive volume, getting huge reach, and your CPM is dirt cheap. When your CPM in the US market is $3 or less, honestly, does it matter whether you can attribute every single person? Probably not.
That said, if you are going to do this, you should definitely measure. Especially when you’re starting out with low traffic, one channel, or no channels - you can immediately see the impact.
It was easier for me at Aithor because traffic was low. For example, we were targeting English-speaking countries, right? We made a video and suddenly we’re seeing payments from Nigeria. We never targeted Nigeria, but somehow they’re converting.
We did a lot of work figuring out how to measure effectiveness. What we landed on: we know conversion rates by country from direct traffic. When we buy a video, we know the creator’s geographic breakdown - say, 50% US, 25% UK, 25% Nigeria. We assume the video will reach roughly the same geographic mix. Then, knowing our direct traffic by country and those conversion rates, we can calculate: what did we actually earn from this influencer?
Host: What about CAC?
Anatoly: That’s the second key metric I used at Aithor: CAC - customer acquisition cost. Total marketing spend divided by new paying subscribers. It’s super straightforward.
When we were scaling and hit half a million in monthly revenue, our CAC was around 35 to 40 bucks, and we were getting roughly 35 to 40 from first payments. So it all made sense.
I think CAC is a really sobering metric. We had a moment where CAC jumped to 75, and we realized we needed to cut something. It seemed like it would be painful, but we got it back down to 40 in two weeks. We just dropped everything that didn’t feel high-quality, cut costs on certain influencers. And boom - two weeks, back to 40.
That’s actually how we got from zero to 1.5 million in monthly revenue - by being ruthless about that metric.
Scaling the Team
Host: You mentioned 400 videos a month. How many people do you need for that?
Anatoly: People are trying to automate this through AI. But right now, we just understood what works and how to scale it. It scales through people.
My approach was that influencer managers owned the entire cycle - finding influencers, negotiating, discussing content, publishing. Everything. The downside is that even a seasoned manager does about six integrations a week with this approach. If you need 60 a week, you need 10 managers. It scales with people.
I had over 30 influencer managers at one point. I built teams, promoted leaders. And I didn’t hire many people with leadership experience. I just watched who stood out and promoted them to leads, helped them become leaders.
We ended up with this big team, and at some point we basically cleared the market and realized we needed interns. Interns are super effective for influencer marketing, honestly. After a month they’re actually working at the same level as experienced people. With influencer marketing you can get up to speed fast.
Host: Thirty people though - that’s basically your own agency.
Anatoly: Yeah, 30 people. That was incredible. We created two teams, Run and Change. Inside those big teams were groups with team leads. There was this moment when I realized there were people I didn’t really know that well anymore. That was an eye-opener - things got serious.
The absolute realization was when people I wasn’t directly managing fired someone. I was like, wow. They came to me saying this person isn’t working out, we need to let them go. That’s when you know things are real.
Hockey Stick Moments
Host: Was there a specific moment when you felt the inflection - like a hockey stick moment?
Anatoly: There were two hockey stick moments, actually. The first was October-November, then the second one was March through May. March, April, May - we were doubling week-over-week. Just straight up. The October-November jump was smaller relative to that, but still significant.
Host: Was that related to seasonality?
Anatoly: Seasonality definitely plays a huge role and affects conversion rate a lot. During those hockey stick moments, conversion always went up. I’d say it’s the combination of virality, the messaging, the traffic appearing, and conversion rising together.
For a learning product, seasonality is huge. Our customers are mostly in the Northern Hemisphere, and education-focused months like October-November and April-May see way higher conversion.
Going Solo: The Vibecoding Journey
Host: What are you working on now? It sounds like you’ve decided to go the solopreneur route.
Anatoly: Yeah, basically I’m doing everything myself now. So it started with Twitter, actually. I kept reading about this vibe of someone building a one-person business. I thought, that’s amazing.
During COVID, I opened a Swift coding course. I tried programming, realized it’s super hard. Then ChatGPT came out. I asked it to write code and I saw something that looked like code. I thought, “Whoa, perfect. Now I’m a solopreneur. This thing will code everything for me.”
Then I started testing and realized something was off, nothing worked. But it looked like code, I couldn’t understand anything.
Then Cursor came out and that’s when I first felt it - I can do everything. I had different ideas and eventually started making my own app. I named it Chiaro, an Italian word.
Host: What’s the app about?
Anatoly: The idea is, when you see something interesting, you wonder what it is. There’s no universal solution that answers all these questions. AI solutions often get it wrong. I figured out how to do it better and realized I could build it with vibecoding tools.
Started building with vibecoding and it’s in the App Store now. It’s not perfect - there are rough edges. I was at Apple’s developer conference and they were talking about tools in Xcode to find where your app slows down, profiling tools. I thought that was cool - first time I heard about it even though my app was already in the store.
Host: Full vibecoding?
Anatoly: Yeah, full vibecoding. I brainstormed the name with Claude. Generated the logo in Midjourney. Discussed the logic with ChatGPT and DeepSeek. I mostly built with Cursor using Anthropic models. Then Claude Code came out and it’s even more efficient than Cursor. Switched to it and I’m really enjoying it. It’s the most powerful thing out there.
They say Claude Code is the product for developers, vibecoding for developers. It works in the terminal. When I first saw it in the terminal I thought it looked complicated, but actually it works better than Cursor because when you don’t understand code, you rarely fix it.
Claude Code sometimes says, “I fixed everything, what’s next?” It literally asks. And moments like when I didn’t understand GitHub, I asked how, and it writes instructions. “How do I upload to Xcode now? How to the web?” It writes step-by-step instructions. I run into errors, bring them back, “Hey, the code spit out errors, look and fix.” And gradually, gradually it all comes together in the App Store.
Host: How long did that take, from start to publishing?
Anatoly: About five months. But it wasn’t every day. I had a day job the whole time, did this in free time. If I did it all over now, I could probably do it in a month. Maybe two weeks full-time.
Now I understand how to improve - my prompting skill got better, my ability to direct the code tool to do what I need. And the models themselves improved too. When I started, there was Claude 3 Sonnet, and it struggled with some things. Now we have Claude 4 Sonnet. It’s getting smarter. Those issues you can’t solve yourself, it suddenly starts solving them.
Vibecoding as a Superpower for Marketers
Anatoly: Overall, I really think vibecoding for marketers is a superpower. There are a lot of marketers who understand business pretty well and understand what people will react to, what’ll be interesting to them, right? Because in marketing we do this constantly. We come up with creatives, offers, landing pages, change headlines.
But a product is just way more complex. Development is needed. You have to think through how it all works. Marketer influence dropped on one hand, but on the other hand, there are no marketers who can program themselves, because it’s a completely different profession.
But simpler apps - not even just the simplest ones, but with persistence, I think you can make medium-complexity apps, add some design, and everything works. Make it look nice and beautiful. That’s totally doable.
Host: What would you recommend for someone who wants to try vibecoding?
Anatoly: You can watch videos, there are different solutions on YouTube where you can see people doing this in real-time. And I’ll say again, read Twitter. There are a lot of creators talking about this and showing it step by step.
But I’d say you just have to try it yourself. There was this viral tweet about what the most popular programming language is now - English, because you don’t need anything else. You can vibecoding in Russian too, it understands it fine.
I’d watch a video and just sit down and start making something. Basic Cursor costs like 20 bucks. You can start with that. You can grab Claude Code, open it and see how it all works, try it out. Or Google AI Studio.
I think vibecoding courses are absolutely useless. Better to understand your own example, because gradually you figure out things. Once you’ve figured out the basics - like how to accept payments in Xcode so everything works, or so events fire in your app - nobody can take that knowledge away from you. Your next app will be faster.
So I think you just have to try and do it. And honestly, it’s such an empowering feeling - wow, it worked, and I did it, I did it myself. That’s incredible.
And here’s the reality: a developer costs about 5,000 dollars a month, right? But Claude Code, the most expensive subscription is 200 dollars. It’s incomparable to developers in any way. So try it, do it. That’s my advice.
Host: That’s cool. Thanks so much for coming.
Anatoly: Thanks for having me. See you.
Transcript edited for clarity. Original interview in Russian.