“And we really rapidly became bottlenecked on other things like our merge queue, which is the get in line to get your change accepted by the system that then deploys into production. We had to completely re-architect it because so much more code was being written and so many more pull requests were being submitted that it just completely blew out the expectations of it.”
Quote · trunk.io
From “Merge fast or merge cheap: Fine-tuning merge queues to handle an increase in PRs from AI agents”, found 2026-10-14.
Quoted in a merge-queue vendor’s post, which does not link the original talk, so this is Trunk’s transcription. Trunk’s own simulation shows the trade-off: batches of 16 cut CI time per PR by 77.5% but merged PRs 54.9% slower.