Article
Managing review when coding agents outpace it
Coding agents can now turn out changes faster than people can properly look at them. This is how I'd work back from that point through the task, the agent's working space and the merge rules so review still means something.
Picture a review queue full of pull requests that all look reasonable, each with tests, a tidy description and an agent transcript. Someone still has to read the requirement, judge the design, look at what the code now does differently and decide whether any of the evidence can be trusted.
Once agents are starting implementation jobs faster than reviewers can finish the ones already waiting, changes sit around, conflict with each other or get thrown away. Reviewers also lose time jumping between unrelated parts of the system. So I'd plan from how many verified changes the team can actually accept, and cap the implementation work to match that.
On X
Got a take on this?
Share it on X. I read the replies.