--- title: How an AI-Native Engineering Team Does Code Review description: AgentField describes rebuilding pull request review around the reality that AI now writes most of its code. The team argues that code review historically... canonical: https://daily.dev/posts/how-an-ai-native-engineering-team-does-code-review-agfjwvirw twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: How an AI-Native Engineering Team Does Code Review | daily.dev og:description: AgentField describes rebuilding pull request review around the reality that AI now writes most of its code. The team argues that code review historically... og:url: https://daily.dev/posts/how-an-ai-native-engineering-team-does-code-review-agfjwvirw og:image: https://api.daily.dev/og/posts/agFjwvIrw.png og:image:alt: How an AI-Native Engineering Team Does Code Review og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # How an AI-Native Engineering Team Does Code Review **[AgentField.ai](https://daily.dev/sources/agentfield)** · 17 min read · 1 upvotes · 1 comments ## Summary AgentField describes rebuilding pull request review around the reality that AI now writes most of its code. The team argues that code review historically bundled four separate jobs—correctness, knowledge transfer, risk visibility, and distributed accountability—that only stayed together because review was slow. With AI writing and checking code, correctness got cheap, knowledge transfer became one-directional, and risk visibility became more critical, while accountability still requires a human name in the audit log. They now treat the PR gate as a 'risk map' rather than a pass/fail verdict, scoring risk across dimensions (security, naming, architectural fit) with adjustable thresholds instead of a single certification. They open-sourced the reviewer system, PR-AF, which spawns per-dimension reviewer agents, runs adversarial self-challenges to cut false positives, and posts inline GitHub risk maps for roughly $0.80 per 500-line PR in LLM costs, alongside their SWE-AF orchestration system for autonomous coding agents. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is PR-AF and how does it review pull requests differently from a typical AI code review bot? PR-AF is an open-source reviewer that spawns separate reviewer agents per risk dimension (such as security, naming consistency, or cross-file interaction) instead of running a fixed checklist, runs them in parallel, has them adversarially challenge their own findings to suppress false positives, and posts the results as an inline GitHub risk map rather than a pass/fail verdict. A deep review of a five-hundred-line pull request costs about eighty cents in LLM calls. _Teams rethinking how AI-generated code gets reviewed can follow tools like this one on daily.dev._ ### Why might treating code review as a pass/fail gate stop making sense when AI writes most of the code? Pull request review historically bundled four jobs together—correctness, knowledge transfer, risk visibility, and distributed accountability—and they only traveled together because review was slow enough for all four to happen in one pass. Once AI writes code at much higher speed, correctness checking gets cheap and parallel, knowledge transfer stops flowing back to the human author, and risk visibility becomes the load-bearing job, while accountability still requires a human name in the audit log. _Engineers redesigning review workflows around AI-written code can track this kind of practice shift on daily.dev._ ## Community discussion Top comments from developers on daily.dev. **@nark3d** · 0 upvotes > Scoring security and naming on the same scale implies you know how they trade off against each other. What happens when a PR scores low on architectural fit but clean everywhere else, does it still get through on the numbers or does someone overrule the threshold? And is the $0.80 per PR the whole cost, or just the reviewers, because the adversarial pass sounds like it runs more than once. Genuinely curious how the thresholds got set. ## Similar posts on daily.dev - [Agentic Code Review](https://daily.dev/posts/agentic-code-review-0ue8fcsa1) · Addy Osmani · 146 upvotes · 14 comments - [Orchestrating AI Code Review at scale](https://daily.dev/posts/orchestrating-ai-code-review-at-scale-re4wh2np4) · Cloudflare · 0 upvotes · 0 comments - [Your old PR comments can now review code](https://daily.dev/posts/your-old-pr-comments-can-now-review-code-fprjbtxni) · Manager.dev · 5 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#github](https://daily.dev/tags/github), [#code-review](https://daily.dev/tags/code-review), [#vibe-coding](https://daily.dev/tags/vibe-coding) [View this post on daily.dev](https://daily.dev/posts/how-an-ai-native-engineering-team-does-code-review-agfjwvirw) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"How an AI-Native Engineering Team Does Code Review","url":"https://daily.dev/posts/how-an-ai-native-engineering-team-does-code-review-agfjwvirw","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-an-ai-native-engineering-team-does-code-review-agfjwvirw"},"datePublished":"2026-10-02T17:45:22.360Z","dateModified":"2026-10-02T18:13:34.505Z","description":"AgentField describes rebuilding pull request review around the reality that AI now writes most of its code. The team argues that code review historically...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/4810a34e065b06987cad19927f3d2492?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/4810a34e065b06987cad19927f3d2492?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"AgentField.ai","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"AgentField.ai","logo":"https://media.daily.dev/image/upload/s--VLFHaedl--/f_auto/v1791063152/squads/33beeb91-1f14-4d9b-9cf8-db7d9f69160a?_a=BAMAMicg0","url":"https://daily.dev/squads/agentfield"},"commentCount":1,"discussionUrl":"https://daily.dev/posts/how-an-ai-native-engineering-team-does-code-review-agfjwvirw","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"keywords":"ai-agents,github,code-review,vibe-coding","timeRequired":"PT17M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"AgentField.ai","item":"https://daily.dev/squads/agentfield"},{"@type":"ListItem","position":3,"name":"How an AI-Native Engineering Team Does Code Review"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/how-an-ai-native-engineering-team-does-code-review-agfjwvirw","comment":[{"@type":"Comment","text":"Scoring security and naming on the same scale implies you know how they trade off against each other. What happens when a PR scores low on architectural fit but clean everywhere else, does it still get through on the numbers or does someone overrule the threshold? And is the $0.80 per PR the whole cost, or just the reviewers, because the adversarial pass sounds like it runs more than once. Genuinely curious how the thresholds got set.","datePublished":"2026-10-11T16:31:35.866Z","url":"https://daily.dev/posts/agFjwvIrw#c-HnWpvHCFC","author":{"@type":"Person","name":"Adam","url":"https://daily.dev/nark3d","image":"https://avatars.githubusercontent.com/u/2162621?v=4"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-an-ai-native-engineering-team-does-code-review-agfjwvirw#faq","mainEntity":[{"@type":"Question","name":"What is PR-AF and how does it review pull requests differently from a typical AI code review bot?","acceptedAnswer":{"@type":"Answer","text":"PR-AF is an open-source reviewer that spawns separate reviewer agents per risk dimension (such as security, naming consistency, or cross-file interaction) instead of running a fixed checklist, runs them in parallel, has them adversarially challenge their own findings to suppress false positives, and posts the results as an inline GitHub risk map rather than a pass/fail verdict. A deep review of a five-hundred-line pull request costs about eighty cents in LLM calls. Teams rethinking how AI-generated code gets reviewed can follow tools like this one on daily.dev."}},{"@type":"Question","name":"Why might treating code review as a pass/fail gate stop making sense when AI writes most of the code?","acceptedAnswer":{"@type":"Answer","text":"Pull request review historically bundled four jobs together—correctness, knowledge transfer, risk visibility, and distributed accountability—and they only traveled together because review was slow enough for all four to happen in one pass. Once AI writes code at much higher speed, correctness checking gets cheap and parallel, knowledge transfer stops flowing back to the human author, and risk visibility becomes the load-bearing job, while accountability still requires a human name in the audit log. Engineers redesigning review workflows around AI-written code can track this kind of practice shift on daily.dev."}}]} ```