Claude and Gemini crush system design and high-level modelling, but the moment you drop into actual function logic, they fall apart. It s like pairing with someone who writes clean diagrams but chaotic code.
And if you ask them to improve it, they vanish into SOLID-theory rabbit holes and over-engineered OOP madness.
These days, since I vibecode more, I have found myself refactoring what the robots are writing ever so often that it has got to a point where I straight up delete function logic and write it myself.

Curious if others are seeing the same?

AI dev tools are evolving crazy fast , every few weeks there s a new must-try for vibe coders.

Some people are building full products with @ChatGPT by OpenAI and @Replit , others swear by @Cursor and @Claude by Anthropic , and a few are mixing @Lovable + @v0 by Vercel + @bolt.new to ship apps in record time.

I ve been refining my own vibe stack lately, trying to find that sweet spot between speed, control, and creativity.
It made me wonder ,what does your setup look like right now?