--- title: The Human Edge: Why Brownfield Codebases Need Mob... description: Two senior developers at Norwegian banking software company SpareBank 1 Utvikling discuss their team's evolution from continuous deployment and pair... canonical: https://daily.dev/posts/the-human-edge-why-brownfield-codebases-need-mob-programming-not-just-ai-vibes-lbfu2or0z twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: The Human Edge: Why Brownfield Codebases Need Mob Programming, Not Just AI Vibes | daily.dev og:description: Two senior developers at Norwegian banking software company SpareBank 1 Utvikling discuss their team's evolution from continuous deployment and pair... og:url: https://daily.dev/posts/the-human-edge-why-brownfield-codebases-need-mob-programming-not-just-ai-vibes-lbfu2or0z og:image: https://api.daily.dev/og/posts/LbfU2oR0Z.png og:image:alt: The Human Edge: Why Brownfield Codebases Need Mob Programming, Not Just AI Vibes 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 Human Edge: Why Brownfield Codebases Need Mob Programming, Not Just AI Vibes **[InfoQ](https://daily.dev/sources/infoq)** · 35 min read · 0 upvotes · 0 comments ## Summary Two senior developers at Norwegian banking software company SpareBank 1 Utvikling discuss their team's evolution from continuous deployment and pair programming to experimenting with Claude Code for four months. They found LLMs excellent for analysis, telemetry dashboards, prototyping, and kickstarting tasks, but poor for generating production code in their 'brownfield' domain, where decades-old backend systems require tacit context the AI lacks. After a Claude-first period from January to May, the team returned to traditional TDD-driven coding with Claude used more as a helper. They also describe how constant mob and pair programming, plus a strict no-outside-pull-requests policy, enables new hires to ship production code on their first day and builds psychological safety and shared domain ownership. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why did a software team stop using Claude Code for writing production code in a legacy banking system? After using Claude first for everything from January to May, a banking software team found it produced code that didn't fit their domain, especially with legacy backend systems containing 50-60 years of history and undocumented quirks. The AI lacked tacit context about API quirks and organizational rules, causing failed integrations even after review, so the team returned to manual TDD-based coding with Claude used mainly for analysis and kickstarting tasks. _Developers weighing where AI coding agents help versus hurt on legacy systems can track real-world experience reports on daily.dev._ ### What tasks are AI coding assistants like Claude actually good for according to developers working on legacy systems? Claude proved most valuable for analyzing telemetry, querying application logs, generating dashboards like Grafana visualizations, kickstarting implementation plans, and rapid front-end prototyping to show clients working mockups within a day or two. It was considered unreliable for writing production code in complex domain logic, since it lacked the tacit knowledge needed to handle quirky legacy backend systems correctly. _Teams deciding how to fit AI coding tools into their workflow can compare real usage patterns on daily.dev._ ### How does constant mob and pair programming help onboard new developers quickly? A banking software team reported that new developers, including juniors and visitors from other teams, had code running in production before lunch on their first day by joining mob programming sessions with roles rotating every 5 to 10 minutes. This approach, combined with narrowing task scope and a no-outside-pull-requests policy, builds psychological safety since newcomers never have to prove themselves alone before contributing. _Engineering leads exploring onboarding practices for fast-moving teams can find similar workflow experiments on daily.dev._ ## Similar posts on daily.dev - [The Earth is not flat](https://daily.dev/posts/the-earth-is-not-flat-p5b0sytuz) · Medium · 0 upvotes · 0 comments - [Mobbing with AI](https://daily.dev/posts/mobbing-with-ai-lscwg8pqi) · Atlassian · 0 upvotes · 0 comments - [Onboarding Claude Code like a new developer: Lessons from 17 years of development](https://daily.dev/posts/onboarding-claude-code-like-a-new-developer-lessons-from-17-years-of-development-dna8a19yg) · Claude · 9 upvotes · 0 comments - [The last generation of coders](https://daily.dev/posts/the-last-generation-of-coders-wxynkly8t) · TechCentral · 1 upvotes · 0 comments --- Tags: [#ai-coding](https://daily.dev/tags/ai-coding), [#claude-code](https://daily.dev/tags/claude-code), [#tdd](https://daily.dev/tags/tdd) [View this post on daily.dev](https://daily.dev/posts/the-human-edge-why-brownfield-codebases-need-mob-programming-not-just-ai-vibes-lbfu2or0z) ```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 Human Edge: Why Brownfield Codebases Need Mob Programming, Not Just AI Vibes","url":"https://daily.dev/posts/the-human-edge-why-brownfield-codebases-need-mob-programming-not-just-ai-vibes-lbfu2or0z","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/the-human-edge-why-brownfield-codebases-need-mob-programming-not-just-ai-vibes-lbfu2or0z"},"datePublished":"2026-08-24T11:00:35.927Z","dateModified":"2026-08-24T11:01:02.158Z","description":"Two senior developers at Norwegian banking software company SpareBank 1 Utvikling discuss their team's evolution from continuous deployment and pair...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/f2938211cfaf4793039106669285c1e3?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/f2938211cfaf4793039106669285c1e3?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"InfoQ","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":"InfoQ","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/afc3bced3e1e4b188dd9127017a60e0c","url":"https://daily.dev/sources/infoq"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/the-human-edge-why-brownfield-codebases-need-mob-programming-not-just-ai-vibes-lbfu2or0z","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-coding,claude-code,tdd","timeRequired":"PT35M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"InfoQ","item":"https://daily.dev/sources/infoq"},{"@type":"ListItem","position":3,"name":"The Human Edge: Why Brownfield Codebases Need Mob Programming, Not Just AI Vibes"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/the-human-edge-why-brownfield-codebases-need-mob-programming-not-just-ai-vibes-lbfu2or0z#faq","mainEntity":[{"@type":"Question","name":"Why did a software team stop using Claude Code for writing production code in a legacy banking system?","acceptedAnswer":{"@type":"Answer","text":"After using Claude first for everything from January to May, a banking software team found it produced code that didn't fit their domain, especially with legacy backend systems containing 50-60 years of history and undocumented quirks. The AI lacked tacit context about API quirks and organizational rules, causing failed integrations even after review, so the team returned to manual TDD-based coding with Claude used mainly for analysis and kickstarting tasks. Developers weighing where AI coding agents help versus hurt on legacy systems can track real-world experience reports on daily.dev."}},{"@type":"Question","name":"What tasks are AI coding assistants like Claude actually good for according to developers working on legacy systems?","acceptedAnswer":{"@type":"Answer","text":"Claude proved most valuable for analyzing telemetry, querying application logs, generating dashboards like Grafana visualizations, kickstarting implementation plans, and rapid front-end prototyping to show clients working mockups within a day or two. It was considered unreliable for writing production code in complex domain logic, since it lacked the tacit knowledge needed to handle quirky legacy backend systems correctly. Teams deciding how to fit AI coding tools into their workflow can compare real usage patterns on daily.dev."}},{"@type":"Question","name":"How does constant mob and pair programming help onboard new developers quickly?","acceptedAnswer":{"@type":"Answer","text":"A banking software team reported that new developers, including juniors and visitors from other teams, had code running in production before lunch on their first day by joining mob programming sessions with roles rotating every 5 to 10 minutes. This approach, combined with narrowing task scope and a no-outside-pull-requests policy, builds psychological safety since newcomers never have to prove themselves alone before contributing. Engineering leads exploring onboarding practices for fast-moving teams can find similar workflow experiments on daily.dev."}}]} ```