Introduction
Meet Frank
Frank is a senior mobile developer.
10+ years shipping apps for companies like Disney Plus. He's built video streaming platforms that millions of people use.
The technical things were never the problem.
But here's what happens when you're really good at building for others: the consulting gigs keep coming. The paychecks are solid.
And that idea you have? It can wait until next quarter.
Except it never does.
Everything changed when AI tools actually got good.
Frank realized he could finally build what he'd been thinking about for years - an AI assistant that actually works. Something that can read your emails, check your calendar, and perform real tasks.
The vision was clear, right?
Why Frank joined the Incubator
Frank needed someone who'd actually built and shipped AI apps to say: "Stop overthinking this. Here's how you solve the auth problem. No, you don't need a database yet. Just launch."
The incubator gave him three things:
1. Direct access to solutions, not theory.
When he got stuck, I showed him my code from a recent project. We debugged it together on a call. Problem solved in 30 minutes instead of weeks.
2. Weekly accountability that actually works.
You can't bullshit your way through a check-in call. Either you shipped something, or you didn't. That pressure matters.
3. Other founders at the same stage.
People building, launching, dealing with App Store rejections, figuring out pricing. Not beginners asking "which framework should I learn?"
As Frank put it: "I don't need motivation because I'm a very motivated person. But I don't have that many developer friends. It's kind of like me talking to my girlfriend about this, she's like, 'Okay,' she doesn't know."
He needed his people.
The incubator's 10-week timeline forced the decision: ship or get left behind.
Problem: too many ideas
"I can make anything," Frank told me on our first call. "But how do I not make everything at the same time__?"
That's the curse. When you're experienced, you see all the possibilities.
And Frank had something most founders dream about: 700 people in his Discord. 300 on a waitlist. People actively asking when this would launch.
But he was stuck in research mode. Classic founder paralysis - not from lack of ideas, but from too many of them.
Then, he showed me his feature list:
- AI voice assistant
- Email integration
- Calendar sync
- Text-to-speech AND speech-to-text
- Multiple MCP servers for Gmail, WhatsApp, Calendar
I stopped him. "What's the ONE thing you want to ship in 10 weeks?"
He kept pivoting between ideas. The AI agent. Email reader. Voice transcription.
We pulled up ASO data together. Searched what people actually type into the App Store.
Turns out, nobody searches for "AI agent that reads my email while I'm driving using MCP servers."
They search for "speech to text," "voice notes," "AI note taker."
We pivoted to a focused V1. Not Jarvis with 15 features.
A voice notes app powered by ElevenLabs that is simple, useful, and shippable.
The Launch: you’ll be shocked to know what users ACTUALLY want
Frank launched to his waitlist.
760 downloads in the first weekend.
He expected people to use it for productivity: email summaries, calendar management, you know, the serious stuff.
Instead, people were using it for:
- Relationship advice
- Horoscopes
- Sports scores
- Language learning
Wait, what? 😅
Turns out, his users in Saudi Arabia were speaking to the AI in Arabic and asking how to say phrases in English. In Hungary, people were shocked it spoke fluent Hungarian. His Discord was lighting up with questions like "Is the AI married?" and "Can it tell me my horoscope?"
Frank's initial reaction: "This is not what I built this for."
My reaction: "Frank, people are telling you exactly what they want. Are you listening?"
This is where most developers quit or force their vision onto users.
Frank didn't.
He talked to UGC creators. Asked what would make people download the app. Their answer?
"Make it feel like FaceTime with an AI friend. Let people share their conversations. That's what'll go viral."
He started experimenting. Added conversation sharing. Tested language learning angles. Explored the horoscope vertical.
"I don't mind that," he told me. "If it can get more mass market appeal, I don't mind. The idea is the initial agent can have other agents - sports, relationships. Then it becomes more useful."
The product evolved based on real user behavior, not his original assumption.
This is the discipline that's hard to teach: killing your darlings, following the data, staying flexible.
What Frank learned in 10 weeks:
One of the best parts of the incubator?
Coffee talks with experts who've actually done this.
We brought in Adam Lyttle: an indie mobile developer who's scaled apps to millions of downloads, and the guys were able to ask anything about their specific situation.
"These little things are really awesome, man," Frank said after. "It's helping me sharpen up everything. I'm getting great feedback from you guys, too."
Direct access matters
"I don't have that many developer friends. What I needed was someone who'd actually built and shipped AI apps. When I got stuck, we debugged it together. That direct access to someone who'd solved these exact problems__? That's what mattered."
On the biggest challenge:
"I can make anything. But how do you not make everything at the same time? How do I do one thing that gets people interested? That's the hard part."
The Real Win
In the first three weeks:
- 760+ downloads
- Active users across three continents
- Multiple iterations based on real feedback
But the numbers aren't the point.
He discovered his users want something different than his original vision, and he adapted.
He's working with UGC creators to find product-market fit.
Most importantly, he has his app out there 💪
If you are interested in becoming one of the next Incubator founders, apply here.
Cheers,
Vadim
P.S. Frank's app is called Loop AI. Check it out - and hey, maybe one day someone will download YOUR app from a P.S. section like this.
