For engineers exploring an unfamiliar codebase, the documentation experience usually goes one of two ways: there's almost none, or there's so much you don't know where to start. Neither is great.
I've been using AI tools to explore new codebases, and it's changed how I learn. Not because it replaces good docs; it doesn't. But it gives you something docs never could: answers tailored to your confusion. You ask the questions bugging you, dig in, go deeper. Follow the thread that's interesting to you, not the one someone else decided to document three years ago.
What surprised me most: this isn't just valuable when docs are thin. Even with thorough documentation, having a conversation about a codebase surfaces insights you wouldn't get from reading alone. It's the difference between a textbook and a tutor who can meet you where you are.
It's personalized onboarding for any codebase, anytime, and one of the most fun ways I've found to learn something new. Pointing an AI tool at an unfamiliar repo and just... asking it questions has become one of my go-to moves. The fog lifts faster than you'd expect.