I treat my AI coding assistant like a brilliant junior dev.
It’s incredibly fast. But it has no context. It doesn’t understand the why behind our architecture.
If I don’t review its code, every small shortcut or mistake becomes a new, bad pattern. Over time, this makes the code messy and unstable.
Our job extends beyond writing code. It’s guarding the integrity of the system.
For new ideas, we let it run wild. An AI assistant is perfect for prototyping.
But in our core codebase, it must follow the rules.
This has forced us to be better at documenting. Style guides, ways of working, and architectural rules are no longer just for humans. They are the playbook we hand to our AI.