NotebookLM has quietly become one of my most-used tools for learning. I use it to break down long documentation, understand unfamiliar concepts, and explore topics without jumping between tabs or losing context. It helps me think through things, not just look them up. Over time, I started wondering what would happen if I used the same approach on something deeply familiar, i.e., my own code. Could it help me reason better, or learn something new? That curiosity led me to try an unusual experiment: giving NotebookLM full access to a codebase and treating it like a human teammate. The results were more interesting than I had expected.