This post discusses the assumption in code reviews that the author of a pull request (PR) understands their own code, and introduces the concept of 'agentic development' which challenges this notion. It argues that understanding is not always guaranteed and explores the implications of this perspective on the code review process.