I was wrong about vibe coding on greenfield projects.
Why setting data models, data flow, and structure before leaning on AI matters more on greenfield work than I thought.
I used to think that vibe coding was good for greenfield projects. I was wrong.
If you don't care about the code and all you want to do is just to test your idea, then it is merely a throw-away PoC not a project. And yes, vibe coding is great for that.
However, as harnesses and models got better over the time, agents started working better on existing codebases. Often times, agents discover existing approaches/code style in the codebase and they start coding accordingly.
I realized that in a greenfield project it is important to set the data models and data flow and general structure of the codebase before handing it off to AI blindly. Otherwise it becomes an unmaintainable mess, and you never want to look at that code again.