I've been building a product at work for 6 months. Every line of code was written by AI. It still isn't finished.
People keep telling me AI makes building software easy. That hasn't been my experience.
AI is very fast at writing code. But typing code was never the slow part. The slow part is deciding what to build, keeping the design coherent across hundreds of prompts, and remembering why you chose something three weeks ago. The AI doesn't carry that for you. It will also happily generate two pieces that quietly contradict each other, then help you debug the mess it made.
Fast code means fast problems. A big share of these 6 months has gone into reviewing, untangling, re-prompting and fixing things that looked fine when they were generated.
I'm not going back to working without it. It's a great tool. But it changed where the work sits for me: less typing, more judgment. Design, review, debugging, trade-offs.
My own project tells the same story. Publlsh, the tool I'm posting this from, is built the same way. I started it in February, about 9 months ago, and it's still not complete.
So no, I don't think AI replaces engineers. If anything, the people who understand how systems fit together get the most out of it.
Anyone else building with mostly AI-generated code? I'd like to hear what caught you off guard.
Top comments (0)