Code review is one of the oldest quality methods in software. As AI agents write more and more of the code it becomes even more important, since an agent can write a lot of code that looks reasonable but is wrong.
We let agents do a first review of every change, of both the quality and the security according to OWASP. Then a person reads it and approves. That is the answer to the question of how we know that the code the agents write holds up.
