After rolling out coding agents, PRs soared and code review became the bottleneck
Swimming_Champion_75 · reddit · 2026-11-02
An engineering lead shares that after rolling out coding agents to most of the team this year, the number of PRs went through the roof while lead time barely moved — writing code got cheaper, but everything now piles up in review, CI, and the two people who actually understand the payments service.
Leadership is talking about building a "software factory": treating the whole delivery process as one measurable pipeline instead of separate agents. The hard question is cost: including tokens, compute, human review hours, and failed/discarded runs, what does an accepted change actually cost?
Tools mentioned:
- Orq AI: tracks human effort per run, token/compute costs, lead time per accepted change, rework, and cost per production outcome by pipeline stage; still in early access and unproven on production repos;
- Factory AI and Devin: focused more on the coding agent itself than on delivery-pipeline economics.
For now, cost per merged agent PR lives in a spreadsheet full of assumptions; the author suspects review hours will be the biggest line item and asks how other teams measure this.
Related event: AI-generated code floods PRs, review becomes the new bottleneck(2 posts)→
More from coding & agent
- DeepSeek design doc: a person can only follow 3-5 agent sessions at once, attention is the real bottleneck — heyshrutimishra · 2026-11-02
- Software fundamentals are key to getting good results from AI coding agents — bendee983 · 2026-11-02
- Open-source LLM regression gate digline hits 1.0: quantifying judge noise before calling a regression — alexpran · 2026-11-02
- PR adds patient retry logic for model capacity errors in OpenAI Codex — bytebot · 2026-11-02
- Perplexity open-sources WANDR: 500 research tasks, 170,495 evidence points, URL-rechecking grader — thisguyknowsai · 2026-11-02
- 4 AIs, 96 questions, 2,671 citations: 74% are blog posts, 99% on companies' own sites — tibo_maker · 2026-11-02