More of an experiment, but thinking as the AI gets better, and assuming Open AI isn't reading things from the API itself, could be used to vet for security or give other kinds of stylistic feedback...<p>... once the AI gets better, might even offer some refactoring tips in its response.