Growing from $0 to $186k MRR in 15 months

After many failed side projects, 30-year-old Jozef Gherman launched StealthGPT a little over a year ago. And it's already bringing in $186k MRR.

He didn't stop there, though. This year, he has launched another three products.

I caught up with him to understand how he does it. 👇

What is your "origin story"?

When I was younger, I had a self-loathing problem, and disliked others in general

My father got me a plaque one day that said, ‘Believe in yourself’. I broke it. He glued it back together and put it on my wall.

Eventually, after high school, I decided to give this idea a try: I would believe in myself. I started treating my life like an experiment. ‘How far can I go?’ And now, I’m able to bulldoze every obstacle in my way.

Turns out, not being the most popular kid in school actually came with some unique advantages.

Not being picked first for the team taught me to pick myself first. Most people go along with the group, but I was forced to create my own groups and lead.

And I became incredibly determined, which is a requirement as an indie hacker. I am far from the most talented, I am far from the smartest, but I am by far the most determined.

Life truly is all about perspective. If every issue you come across is one you believe you can solve with time and resources, your issues melt away like snow on a sunny day.

Why did you start indie hacking?

I’m new to Indie hacking. I started and launched StealthGPT a little over a year ago — February of 2023. I knew it was a good idea, but I did not anticipate it becoming a run-away success.

I only picked up coding back in 2020 to learn how to build old-school RuneScape bots. It was mostly out of a sense of nostalgia — I used to spend days playing Runescape, often losing track of time. Plus, farming in-game items and currency through bots seemed like it might make a bit of money.

So I watched YouTube tutorials and took the famous Harvard CS50 class. After ChatGPT was launched, I used that as a mentor as well. AI is a phenomenal teacher!

Funny thing is that Java was so complicated that I never even figured out how to do it. But it was highly educational — it was fun, so it kept my focus. Learning through failure and eventual success is still the model I use today.

And I think I’ve done pretty well with my development endeavors since then!

What are you working on?

StealthGPT is at $186k MRR. It is the world’s first undetectable AI service. It is subscription-based with tiered pricing that offers varying levels of access. Each plan offers an increasing number of “Stealth Word” limits.

Our business plans just recently became a big center of our strategy. We launched them a few months ago and, while it only makes up 3% of our revenue right now, it's our fastest growing segment.

I’ve also started three other products this year. That might seem a little crazy, but it’s how we experiment with new AI-driven use cases and see if they’re desirable by customers.

• AIBypass launched in January and it’s at $3.2k MRR. It’s an undetectable AI solution, but it has different use cases than StealthGPT.

• RapidResume launched in February and is at $570 MRR. It uses AI to create resumes.

• And AIDetect launched in March. It’s an AI detection solution. It’s at $400 MRR.

Creating these niche micro-products allows us to ship quickly, gather feedback and data, and analyze. Some of the features then make their way back into StealthGPT later on.

How did you come up with the idea?

I have been intrigued by AI from a philosophical and technological perspective since before I can remember. After ChatGPT’s incredible rise, I started doing research into privacy and AI, since I believed the technology would become incredibly invasive.

I noticed that there was little to no work being done in this field, so I decided to plant my flag there. My first idea, StealthGPT, was based on my theory that as AI-generated text usage became evermore prevalent, so too would AI detection systems.

Not a week went by when I saw a viral video about GPTZero, one of the first AI Detectors. I knew at that moment that I needed to make StealthGPT.

Predicting the future requires pattern recognition, and the more data you collect, the more accurately you can see patterns and make predictions. If you see something others don’t, and you have done the research, don’t follow their lead.

How did you build the MVP?

Like any good indie hacker, I started broke. I started by personally loaning $250 to the business. Shout out to Vercel, for having a very generous free tier that allowed me to incubate the business.

I built it using NextJS/React, Vercel, Tailwind, Planetscale, Resend (Notifications), and Clerk (auth).

At the time, I was looking for work while taking consulting jobs and building the prototype in my evenings.

Even though StealthGPT was my main focus, those additional consulting and contracting roles kept the lights on in the early stages. It took Stealth about six months before we made enough revenue to both keep expanding our ads and start paying me

If you want the time to do something, you make the time for it. I don’t believe in excuses. People spend way too much time talking about what they want to do. There are doers and sayers; the former are rare but far more successful.

Subscribe for more how-tos, roundtables, and interviews with people in the thick of it.

How did you grow StealthGPT?

Our app's core function, the AI-detection “Bypasser,” quickly garnered attention from a few tech podcasts and blogs. These mentions boosted our SEO ranking for key terms like “undetectable AI,” driving our first few dozen subscribers and bringing in $2k to $3k in monthly revenue.

It's one of the benefits of working on something that people want to talk about.

While organic growth was promising, though, it wasn't enough. We invested a significant portion of our revenue into user-generated content from small, trustworthy creators on TikTok and Instagram.

We selected them based on engagement metrics and audience alignment. It was challenging to find the right creators, but once we found someone who resonated with their audience and respected our brand, we formed lasting partnerships. Some creators have been so successful that we practically put them on payroll.

Initially, 70% of our marketing budget went there. The rest went to ads. This unlocked another $10k MRR.

How did you make paid ads work for you?

Now, the ratio has flipped, with 80% of our marketing budget being spent on Google, Meta, and X ads, and 20% on creators.

We spend roughly half our revenue on ads — so $93k/mo. Our payroll only constitutes roughly 15-20% of revenue, so even when spending another 65% to 75% of our total revenue on marketing, we remain profitable.

I started on Google Ads with practically zero knowledge of the platform. I couldn't find any helpful tutorials, so I learned through trial by fire.

Our process now consists of targeting the same keywords and search queries that we know are used by our target audience, and utilizing the ads that consistently get us results. I adjust the budget on each campaign here and there and, in many cases, I let Google’s algorithm do its magic.

Getting accurate conversion tracking data from our site was both the trickiest and most rewarding thing we did. It was the ultimate key to optimizing our ads.

How did you optimize your ads so effectively?

Accurate cost per conversion data is everything. It is the backbone of our spending strategy.

To get accurate conversion data, we set up conversion events between our site and Google. A lot of people skip this part.

So, when Google sends us a user and they successfully check out, we fire an event to google to let them know a conversion happened. You need this in place to get good conversion data from google and get your CPA down. 

I know the cost per conversion that we need to target to be profitable with new customers in month one, month two, month three etc. With that knowledge, I adjust accordingly.

And this approach led to exponential growth, with almost every month seeing 50% MoM growth.

What other factors drive your growth?

Ultimately, our core philosophy is what drives our growth the most: Solve a problem, solve it better than anyone else, and solve it in a way that provides such a wide and tangible advantage to your audience that they’d be stupid to be using any other product.

This brings me back to our ads. One interesting campaign was when Turnitin, the largest AI-detector in America, released a new detection model. Ours was the only AI that could consistently beat it.

This fact became the centerpiece of our largest ad campaign, and it skyrocketed our revenue.

What mistakes did you make along the way?

I have this mental ability to run a million miles a minute. Operating at that level makes you far more prone to mistakes, but I learn best through trial by fire, and my battle scars serve as building blocks to elevate myself further.

One key mistake was offering a free tier that was too generous. It was effective in growing our credibility when we were relatively unknown, but it became a source of massive lost revenue when our popularity grew. This became evident when we ran A/B tests — when the number of free users was too high, our weekly revenue would fall dramatically.

Eliminating the free tier altogether gave us a 50% revenue spike that month, and reduced the number of support emails we were getting as well.

This strategy won’t work for every market, some products benefit hugely from a free tier, but it’s well worth evaluating if your free tier is too generous and if your users would be more likely to click “Buy” if there was more incentive.

How did you find the time to launch three new products this year?

The key to juggling four projects is picking your battles. And when you choose which battle to focus on, focus completely on that one thing.

Don’t let anything pull you away, distract you, or prevent you from shipping that feature — even your other projects. Barring a critical failure, they can wait.

So StealthGPT would get sidelined for a month or two while we sprinted on a new idea. Then we'd get back to StealthGPT for a few months before jumping into another project.

Another big factor was building a team. While it was fun to be a one-man-show in the early days, I realized that this project was in serious jeopardy if I didn't expand. Now, we have a team of seven and I get to zoom out from the day-to-day work and focus on high-level strategy and prototyping of new features without fear of site maintenance and support falling off a cliff.

Were you surprised by your runaway success?

I worked on many side projects before this and none of them went anywhere. I often wanted to give up entirely and succumb to working a normal job. That boring, easy, wide-open way. Thank God I didn’t.

At the end of the day, though, I'm not surprised I ended up here. I always knew I’d prove everyone wrong. I believed in my heart that my prayers and ambitions would ultimately be fulfilled.

And I’m just as sure now that we’re going to take this company to the next level.

What does the future hold for StealthGPT?

I am a very ambitious person. I want to keep building quality SaaS products and the next great American Dynasty.

I am supremely confident in myself and my team. We are always kicking around new ideas and it’s just a matter of time before we come up with several more big winners, so keep us on your radar. Lot’s more to come!

You can follow along X, LinkedIn, and my personal site. And check out StealthGPT.

Any parting advice?

Give it your all and go with your gut. You’re either earning or learning. If you should fail, figure out why and get back into action. I failed so many times, and sometimes spectacularly, before I succeeded.

And remember that no one knows anything. Be careful of those who pretend that they do.

Subscribe to the Boot's Trap series