i’m experimenting with an ai “rubber duck” idea that helps engineers think through bugs and design decisions by asking questions instead of spitting out code. this is an early prototype and i’m trying to validate whether ai tools are hurting code understanding.

most ai software engineering assistants like cursor and claude code focus on the implementation details/results but don’t help teams upskill or build in a reliable manner.

would love to get any feedback and thoughts!