As AI-generated code significantly increases pull request sizes and creates severe review bottlenecks, traditional code review is becoming unsustainable. Rachel Laycock argues that instead of relying on mandatory pull requests to catch bugs, share knowledge, and align architecture at the end of a project, teams should "shift judgment left."
By moving these critical conversations to the beginning of the development process—using practices like pair programming, collaborative design sessions, automated testing, and automated linting—teams can collaborate and transfer knowledge in real time. Ultimately, code reviews should be reserved only for exceptional situations, such as high-risk security changes or major architectural updates, rather than serving as a blanket ceremony for every line of code.