Curious to hear how everyone is dealing with the flood of PRs in their teams now that we all arrived in the AI era and dust settled a bit. From review process (humans + agents, when/which) to e2e testing (like having scenarios as markdown that agents test?), where did you land and what were the hard lessons learned?