Blog

I was watching yet another AI demo where someone typed 'build me a todo app' and a beautiful, fully functional application appeared in seconds. Clean design. Smooth animations. Real database. The audience clapped.

And I thought: what if... it was bad?

Not bad by accident — we already have that; it's called 'every app I've ever shipped.' Bad on purpose. Intentionally, gloriously, hilariously bad.

So I built Hatable in a weekend. The first app it generated was a login page where the password field was visible and the username field was hidden. It was the most honest piece of software I'd ever seen.

That's it. That's the origin story. There's no deeper meaning. We're not 'disrupting' anything. We're just making bad apps and having a good time.

If you've read this far, I respect your commitment to content that adds nothing to your life. Welcome to Hatable.

We have a problem. Our AI is too smart.

We ask it to generate a terrible dashboard. It generates a clean, well-organized dashboard with proper data visualization. We say 'make it worse.' It adds a dark mode toggle. That's not worse, that's a feature.

We've tried everything. We've prompted it with 'you are a bad developer.' It interpreted this as 'you are a developer who is going through a bad day' and generated an app with a built-in meditation timer.

We told it to 'ignore best practices.' It responded by writing the cleanest code we've ever seen, just without comments. That's not ignoring best practices, that's a lifestyle choice.

Eventually we found the sweet spot: ask it to build something while simultaneously contradicting every requirement. 'Build a minimal dashboard with 47 widgets. Make it fast but load a 50MB hero image. Use modern design but make it look like it was built in 1997.'

The AI doesn't know what to do. It panics. The output is chaos. It's beautiful.

If any AI researchers are reading this: your model is too good and it's ruining our business.

On December 19th, someone submitted Hatable to Product Hunt. It wasn't us. We don't know who did it. The submission description was just 'lol' and a link.

By morning, we were #1 Product of the Day. 400 upvotes. Comments like 'this is the most honest startup I've ever seen' and 'I can't tell if this is satire or just a really bad product.'

It's both. It's always been both.

Our servers couldn't handle the traffic. This was embarrassing for a product that claims to have 99.1% downtime — the actual downtime was unplanned and much less funny.

We've since recovered. Traffic is back to our usual 3 visitors per day (two are bots). All is well.

Dear users,

We're sorry.

We're sorry for the apps. We're sorry for the downtime. We're sorry that the gallery exists and that you can see what other people have made. We're sorry that the AI sometimes generates something that's accidentally good and you get your hopes up.

We're sorry that the fire animation uses 40% of your CPU. We're sorry about the cookie that won't die. We're sorry that our privacy policy is longer than our actual codebase.

We're sorry that you read this entire blog post expecting a punchline. This was the punchline.

With regret, The Hatable Team

Every quarter, we create a roadmap. It makes us feel productive. Then we never look at it again. Here's Q1 2026's roadmap, published for full transparency:

Q1 Goals: • Make the AI 20% worse (stretch goal: 40% worse) • Add multiplayer mode where two people can ruin an app together in real-time • Launch mobile app (that barely works on mobile) • Achieve profitability (lol) • Hire someone (anyone) • Fix the bug where apps occasionally work correctly

Current progress: We've accomplished none of this. We did add a changelog page though, which wasn't on the roadmap. Off-roadmap work is the only work we do.

Q2 Preview: We haven't thought about it. Q2 is future-us's problem. Future-us is going to be so mad at present-us.

If you're an investor reading this, please note that 'roadmap' is a strong word for what this is. It's more of a 'road suggestion.' A 'road vibe.'