I built an app with AI. The code wasn't the hard part.

I’ve been building my app mostly with AI coding tools, and the experience has been pretty weird.

At first, it feels almost like cheating.

You describe a feature → AI writes it → you test it → repeat.

But once the app gets bigger, the problems change.

Suddenly you’re dealing with:

• AI changing things you didn’t ask it to touch

• features that work individually but break together

• inconsistent UI decisions

• debugging AI-generated code you barely remember approving

• constantly re-explaining how the project works

The biggest lesson for me:

Vibecoding makes writing code cheaper. It doesn't make making good software easy.

I’m curious what other people have experienced.

What was the moment when your vibe-coded project stopped feeling easy?

Was it complexity, context loss, debugging, design, or something else?