I’ve been seeing a lot of people proudly shipping “vibe-coded” websites lately…
And honestly, I get the appeal.
You describe what you want, AI generates the code, things go live fast. No deep engineering, no long dev cycles.
But the more I look at these projects, the more a pattern shows up:
They work… until they don’t.
Some common issues I keep noticing:
Code that looks fine but breaks when traffic increases.
No real structure for scaling or future updates.
Security gaps no one thought about.
Performance that drops the moment real users show up.
And the biggest one → no one fully understands the codebase.
It feels like we’ve optimized for speed of creation, but ignored stability and ownership.
I’m not against AI-generated code. But I do think most people underestimate what happens after the first version goes live.
Curious how others here are thinking about this: Have you shipped something using vibe coding? How did it hold up later?
David Warner
This is a very valid concern and something many developers are starting to realize. Vibe coding works well for quick MVPs, but scaling to 10,000 users often exposes issues in structure, security, and performance. Without a clear understanding of the codebase, long-term maintenance becomes difficult. I’ve seen similar challenges even in platforms like online education where stability becomes critical as traffic grows. AI is powerful, but strong engineering practices are still essential for building reliable and scalable products.