How the Claude Code team does code review / claude.dev Blog

How do you review code at the volume agents produce it? On the Claude Code team, once we had high coverage and deep reviews, we started optimizing for the minimum amount of human review time that doesn’t raise reverts or incidents.

Read in full here:

Working on aicost.tools, the line that stuck with me is “the minimum amount of human review time that doesn’t raise reverts or incidents”. That feels like the honest metric: once agents write most of the code, the expensive part is no longer the tokens, it’s the hour a senior spends reading a diff. Has anyone here actually measured review time per PR before and after bringing agents in?