From Frustrated Prototyping to AI Game Creation

1. The Problem

I've been building software for about eight years now. Web apps, backend systems, the usual stuff.

At some point, I started playing around with game ideas. Not seriously—just personal projects, weekend experiments. But every time I'd get excited about a concept, I'd hit the same wall.

Setting up an engine. Debugging physics. Writing the same boilerplate code I'd written a hundred times before.

By the time I got something playable, the excitement had faded. The gap between "I have an idea" and "I have something I can actually show someone" was just too big.

2. Getting Started

I started thinking about this problem more seriously last year. What if AI could handle the technical parts?

The idea behind SoonLab is straightforward: let people describe a game concept and get something playable back. No engine setup. No coding required. Just idea to prototype.

I didn't start with a grand vision. I just wanted to see if it was even possible.

The first version was rough. I mean, really rough. It could barely handle simple prompts and the output was... let's call it "abstract." But it worked. People could type something and something would come back.

That was enough to keep going.

3. Building in Public

I spent two weeks lurking on Indie Hackers before I could post here. Commenting, reading, trying to understand the community.

What I found was refreshing. No hype. No "growth hacking" buzzwords. Just people sharing honest stories about building things.

So I figured I'd do the same.

The development has been iterative. I launch, watch how people use it, gather feedback, iterate. Repeat. Some features I thought were important got ignored. Some things users asked for turned out to matter more than I expected.

One example: I spent a week building a "game customization" panel. Sliders for difficulty, visual themes, gameplay parameters. Users didn't touch it.

What they wanted was simpler: just describe something and see it work. Less friction, faster results.

4. Current Numbers

I want to be honest about where things stand.

Daily active users: ~200
Total revenue: ~$20
Months in development: Still counting

This is very early. I'm not optimizing for monetization yet. The focus right now is purely on learning: Are people actually creating games? Are they having fun? Are they coming back?

The numbers are small, but they're real. And people are using it.

5. Lessons Learned

A few things have become clear:

Instant feedback matters. If users can't get something playable within a few minutes, they leave. I redesigned the onboarding flow three times to get time-to-first-game down.

Users don't want options—they want outcomes. They want to describe something and see it work. Customization can come later, if at all.

Early users are collaborators. Watching how people actually use the product reveals things that user interviews never catch. I've changed direction based on observing behavior, not surveys.

Build in public when you can. Sharing the journey—even when there's not much to show—creates accountability and connection.

6. What's Next

Continuing to iterate. More templates. Better AI responses. Smoother workflows.

I'm curious where this goes. At what point does AI-assisted creation become AI-driven creation?

For now, I'm just building something that didn't exist before, one iteration at a time.

Your Experience

Has anyone else worked on AI creative tools? What did you learn?

For those who've tried similar products—what actually matters to you?

Would love to hear from others on similar journeys.

Check out SoonLab if you're curious.