I finally got a few small wins under my belt with AI coding, but they were both with small projects.
Context is key:
- Smaller files
- More comments
- Fewer dependencies
Also, the smaller the change, the more likely it is to succeed.
But how can I scale this up to larger, more complex projects? Existing projects with massive files, zero comments, and lots of obscure dependencies…
In the meantime, creating small tools for small tasks seems to be an easy win.