Out of the box yes this is true, but:

1. Custom instructions go a long way
2. Review the code yourself, tell it what to fix, and it will fix it. For me it often takes like 5 rounds of this before the code is fully polished to the point I’m proud of it. And you know what? It’s still MULTIPLE times faster than typing code by hand. And at least for me, the quality is higher, because I have 12 different agents that review the work too and they catch additional issues that even I missed.

If you or others ship shitty code don’t blame the LLM, the issue is entirely the engineer using it wrong

Skill issue tbh

It means that you created the code via natural language instructions to an LLM as opposed to typing by hand. You still need to verify its work, review and understand the code to at least some extent, and go through many iterations before it’s production ready.

Some people create vibeslop without looking at the code at all which is fine for a throwaway hobby project or personal tools but it will very quickly turn into unmaintainable garbage if you don’t guide the architecture, redirect away from bad approaches, review the code carefully and make the LLM fix any issues, ensure proper testing, and periodically refactor and improve code quality. If that sounds like hard work then that’s because it is, doing a good job with agentic accelerated engineering is hard work, but you can still build 2x-5x faster at the same quality level than if you typed it by hand. And that’s with current models, they’re only getting better

Fwiw that’s not what vibe coding means

I like AI