use AI for code review, having another AI (or the same AI in a
new session) audit code. It can examine every line carefully,
never tires, never misses.
"never"? Absolutely roflcoptering over here.
If you've never had an LLM code reviewer hallucinate, you're either not reading the output or you're just lying.