I turn sentences into apps. Here's what it cost to get here.
The honest founder story behind AppX — the doubt, the pivots, the nights spent fixing the unglamorous two seconds at the end of a forty-step pipeline. Not the architecture. The human part.
AppX team ·
I build a thing that turns a sentence into an app. You type "make me a habit tracker with streaks," and about thirty seconds later your phone is holding a real, running app you can tap around in. People see that and assume the hard part was the AI. The hard part was everything around the AI — and most of it happened on nights spent alone in front of a glowing screen.
This isn't the architecture story. I've written that one already, with the version numbers and the things we tore down. This is the other story. The one about what it actually costs to sit alone in a room and decide, over and over, that the thing you're building is worth one more day.
The idea was embarrassingly simple
You should be able to describe an app and get the app. That's it. That was the whole thesis.
It sounds obvious now, in 2026, when half the internet is "describe X and get X." It did not sound obvious when I started. It sounded like the kind of thing you say at a dinner and everyone nods politely and changes the subject. The gap between "wouldn't it be cool if" and "here is a working product people pay for" is where almost everything dies. Most ideas never cross it. Not because they're bad — because crossing it is unglamorous, and lonely, and takes far longer than the burst of excitement that produced the idea.
I crossed it slowly. Badly, at first.
I built the wrong thing first, on purpose, without realizing
The first version of AppX didn't make mobile apps. It made web apps.
I'm not proud of why. I built web first because web is easier. The tooling is mature, every AI model on earth has read a million React tutorials, and you can put a result on a URL in seconds. I told myself this was strategy. It wasn't strategy. It was me taking the path that hurt less and dressing it up as a plan.
The demo looked great. The screenshots looked like a real product. And then I put it in front of the people who had actually asked me for an app builder, and every single one of them said some version of the same sentence:
"Cool — but can I see it on my phone? Like, actually on my phone."
Not a mobile-shaped web page. Not a responsive layout. Their app, in their hand, behaving like an app. I had spent a month shipping the easy version and calling it progress, and they saw through it in one question.
That's the first thing nobody tells you about building alone: there's no one in the room to catch you choosing comfort over truth. You have to be the person who notices, the person who's disappointed by it, and the person who decides to do it the hard way anyway. All three are you. On the same night.
The part where I wanted to quit
I'll be specific, because vague founder-struggle posts are useless.
There was a stretch where the preview — the whole point, the moment where your idea shows up on your phone and works — would just... not show up. The app would generate. The code was right. And the screen would sit there, spinning, blank, lying to the user that something was happening when nothing was.
The cruelty of it is in where the failure sits. Everything impressive already worked:
flowchart LR
A["Understand the idea"] --> B["Plan the screens"]
B --> C["Write the app · 40+ steps"]
C --> D["Bundle & deliver"]
D --> E["Show it on the phone"]
C -.->|"all the hard, clever work"| C
D -->|"failed silently here"| X(("blank<br/>spinner"))
style X fill:#3a1414,stroke:#c0392b,color:#f7d7d7
style E fill:#13241a,stroke:#2ecc71,color:#d7f7e3
The AI wrote a genuinely good app. The user never saw it, because the last two seconds of a forty-step pipeline failed quietly. All that work, invisible, behind a spinner.
I want to tell you I had a breakthrough and it was euphoric. It wasn't. It was the opposite. It was me, methodically, refusing to accept "it works on my machine," reading the actual evidence of what users hit instead of what I hoped they hit, and fixing the unglamorous last two seconds that everyone else would have shrugged at because the cool part already worked.
That's most of the job, honestly. The cool part working is table stakes. The product is the boring two seconds at the end that you have to care about more than is reasonable.
What solo actually feels like
People romanticize the solo founder. Let me de-romanticize it.
Solo means when something breaks at 1 a.m., the person who fixes it is you, and the person who decides whether it's even worth fixing is also you, and there's no second opinion, and you've been staring at it long enough that your judgment is shot. Solo means you are simultaneously the optimist who started this and the realist who has to look at the numbers. Those two people argue. Constantly. In your head. On a Tuesday.
I run this lean, on a tiny budget, with a discipline about cost that comes from there being exactly one wallet and it's mine. Every fraction of a cent of AI usage per generation is a number I feel. That constraint sounds like a limitation. It's actually the best product designer I have. When you can't afford to be wasteful, you're forced to be good. You can't paper over a sloppy system with more compute. You have to make the system right.
What kept me going (it wasn't motivation)
Motivation is a liar. It shows up for the demo and vanishes for the maintenance. If I'd run on motivation I'd have quit during the blank-screen month.
What kept me going was smaller and more durable: every so often, someone who could not code would type a sentence, and a real app would land on their phone, and they'd get this specific look on their face. Not "neat tech." More like — recognition. Like they'd just discovered that the thing they'd carried around in their head for years was actually allowed to exist. That they didn't need permission, or a technical co-founder, or six months and a loan.
I'm chasing that look. That's the whole engine. The architecture, the cost discipline, the nights — all of it is in service of putting that look on more faces.
If you're standing at the same gap
If you have an idea and you're staring at the gap between "wouldn't it be cool" and "it exists," here's the only thing I actually know:
The gap is not crossed by a burst. It's crossed by refusing to stop on a hundred ordinary days, most of which feel like nothing is happening. You will build the easy wrong thing first. You'll defend it. Then you'll have to be honest enough to throw it away. You'll lose nights to problems that aren't even the impressive part. And occasionally — not often, but occasionally — you'll see the look on someone's face, and it'll be enough to buy one more day.
I turn sentences into apps now. It works. People use it. And it cost exactly what it was supposed to cost: every ordinary day I refused to quit.
If you've got the sentence, I built the rest. Come describe your app and watch it land on your phone. That's the look I'm talking about. I'd like to see yours.
— Ilyos, building AppX. Follow the journey @ilyos_ceo.