With the way memory systems work, I can see the value in having a different person's AI conduct the review as that AI's 'memory' is going to have a slightly different perspective aligned with the developer piloting it