We have the same issue and now spending more time/energy on optimizing the code review agent.

This might lead to more automatisation and not less if it works. After all we might extract everything critical to us.

We also now expect juniors to review the other ones code first.