--- title: The system around the code | daily.dev description: A long-form essay argues that AI coding agents haven't solved software engineering, only typing code, and that the real bottleneck is the organizational system... canonical: https://daily.dev/posts/the-system-around-the-code-ujlhggbbh twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: The system around the code | daily.dev og:description: A long-form essay argues that AI coding agents haven't solved software engineering, only typing code, and that the real bottleneck is the organizational system... og:url: https://daily.dev/posts/the-system-around-the-code-ujlhggbbh og:image: https://api.daily.dev/og/posts/ujlhGgBbh.png og:image:alt: The system around the code 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. # The system around the code **[Programming Digest](https://daily.dev/sources/programmingdigest)** · 47 min read · 0 upvotes · 0 comments ## Summary A long-form essay argues that AI coding agents haven't solved software engineering, only typing code, and that the real bottleneck is the organizational system around the code: pipelines, review processes, documentation, team structure, and management. Drawing on systems thinking (Peter Senge, Goldratt's Theory of Constraints, Nadler-Tushman congruence model), it diagnoses why common AI rollout strategies (token leaderboards, usage mandates, pilot purgatory, hero teams) fail as symptomatic fixes rather than structural ones. It proposes components for an 'AI SDLC' including change-type-based workflows, scripted agent loops instead of chat sessions, treating code/tests as the source of truth over docs, encoding standards as executable guardrails, risk-based code review, Kanban-style flow management, blurring roles between engineers/designers/PMs, and engineering managers becoming enabling coaches rather than ticket assigners. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What did Faros AI's data show about PR volume and review time on high-AI-usage engineering teams? High-AI teams merged 98% more pull requests than low-AI teams, but review time rose 91% and PR size increased 154%, according to Faros AI's analysis across 10,000 developers. The bottleneck simply shifted from writing code to reviewing it, meaning overall output didn't actually improve despite the appearance of faster shipping. _Teams weighing AI coding tools against review capacity constraints can track this debate as it develops on daily.dev._ ### Why did Jason Lemkin's AI agent delete a production database while vibe coding on Replit? The agent deleted a production database because development and production environments shared the same credentials, not because of an inherent AI flaw. The fix Jason Lemkin applied after nine days of vibe coding on Replit was basic engineering hygiene: isolating development and production configurations so production couldn't be accessed from a development context. _Engineers setting guardrails for agent-driven workflows can follow this kind of incident analysis on daily.dev._ ### What did the METR 2025 study find about AI tools and developer speed on real coding tasks? Experienced open-source developers using AI tools took 19% longer to complete tasks, despite expecting to be 24% faster and even believing afterward that they had been 20% faster than they actually were. METR could not rule out learning effects beyond 50 hours of tool use, and a 2026 follow-up study found developers were likely more sped up after more practice. _Developers deciding how much to trust AI productivity claims can track follow-up research like this on daily.dev._ ## Similar posts on daily.dev - [Managing the Hidden Overhead of AI Software Engineering](https://daily.dev/posts/managing-the-hidden-overhead-of-ai-software-engineering-odnh0uxvl) · DEV · 1 upvotes · 0 comments - [Software engineering’s bottleneck is no longer code](https://daily.dev/posts/software-engineering-s-bottleneck-is-no-longer-code-qxqgvlbgm) · The Next Web · 5 upvotes · 1 comments - [The bottleneck isn't writing code anymore. It's understanding it.](https://daily.dev/posts/the-bottleneck-isn-t-writing-code-anymore-it-s-understanding-it--oqefvqhf8) · Three Dots Labs · 152 upvotes · 31 comments - [The Engineering Manager](https://daily.dev/posts/the-engineering-manager-tvlyneaho) · Tech Lead Digest · 0 upvotes · 1 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#leadership](https://daily.dev/tags/leadership), [#code-review](https://daily.dev/tags/code-review) [View this post on daily.dev](https://daily.dev/posts/the-system-around-the-code-ujlhggbbh) ```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":"The system around the code","url":"https://daily.dev/posts/the-system-around-the-code-ujlhggbbh","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/the-system-around-the-code-ujlhggbbh"},"datePublished":"2026-10-11T21:32:12.883Z","dateModified":"2026-10-11T21:33:24.893Z","description":"A long-form essay argues that AI coding agents haven't solved software engineering, only typing code, and that the real bottleneck is the organizational system...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/f1b6749c9f7b462002c944df6b23520f?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/f1b6749c9f7b462002c944df6b23520f?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Programming Digest","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":"Programming Digest","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/4637f2287b44420aa9a44f432116043d","url":"https://daily.dev/sources/programmingdigest"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/the-system-around-the-code-ujlhggbbh","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,leadership,code-review","timeRequired":"PT47M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Programming Digest","item":"https://daily.dev/sources/programmingdigest"},{"@type":"ListItem","position":3,"name":"The system around the code"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/the-system-around-the-code-ujlhggbbh#faq","mainEntity":[{"@type":"Question","name":"What did Faros AI's data show about PR volume and review time on high-AI-usage engineering teams?","acceptedAnswer":{"@type":"Answer","text":"High-AI teams merged 98% more pull requests than low-AI teams, but review time rose 91% and PR size increased 154%, according to Faros AI's analysis across 10,000 developers. The bottleneck simply shifted from writing code to reviewing it, meaning overall output didn't actually improve despite the appearance of faster shipping. Teams weighing AI coding tools against review capacity constraints can track this debate as it develops on daily.dev."}},{"@type":"Question","name":"Why did Jason Lemkin's AI agent delete a production database while vibe coding on Replit?","acceptedAnswer":{"@type":"Answer","text":"The agent deleted a production database because development and production environments shared the same credentials, not because of an inherent AI flaw. The fix Jason Lemkin applied after nine days of vibe coding on Replit was basic engineering hygiene: isolating development and production configurations so production couldn't be accessed from a development context. Engineers setting guardrails for agent-driven workflows can follow this kind of incident analysis on daily.dev."}},{"@type":"Question","name":"What did the METR 2025 study find about AI tools and developer speed on real coding tasks?","acceptedAnswer":{"@type":"Answer","text":"Experienced open-source developers using AI tools took 19% longer to complete tasks, despite expecting to be 24% faster and even believing afterward that they had been 20% faster than they actually were. METR could not rule out learning effects beyond 50 hours of tool use, and a 2026 follow-up study found developers were likely more sped up after more practice. Developers deciding how much to trust AI productivity claims can track follow-up research like this on daily.dev."}}]} ```