--- title: What AI Secretly Changed About My Engineering | daily.dev description: After hundreds of hours using AI coding tools like Claude, Copilot, and JetBrains Junie on real production services, the author found that AI didn't just make... canonical: https://daily.dev/posts/what-ai-secretly-changed-about-my-engineering-71x5m4e3y twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: What AI Secretly Changed About My Engineering | daily.dev og:description: After hundreds of hours using AI coding tools like Claude, Copilot, and JetBrains Junie on real production services, the author found that AI didn't just make... og:url: https://daily.dev/posts/what-ai-secretly-changed-about-my-engineering-71x5m4e3y og:image: https://api.daily.dev/og/posts/71X5M4E3Y.png og:image:alt: What AI Secretly Changed About My Engineering 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. # What AI Secretly Changed About My Engineering **[iO tech\_hub](https://daily.dev/sources/iotechhub)** · 15 min read · 27 upvotes · 3 comments ## Summary After hundreds of hours using AI coding tools like Claude, Copilot, and JetBrains Junie on real production services, the author found that AI didn't just make coding faster — it exposed weak engineering habits. Key lessons include: writing clear specifications before requesting code, treating AI context as a finite resource by keeping tasks small and focused, always reviewing AI-generated plans before implementation, avoiding having the same AI review its own output, and maintaining strict diff size discipline. The author advocates for spec-driven development as the core workflow shift, arguing that the habits AI rewards — clear requirements, small changes, independent review, meaningful tests — are simply good engineering practices that outlast any specific tool. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why does using AI coding assistants like Claude or Copilot make ambiguous requirements more costly? Vague task descriptions let AI fill gaps with plausible-sounding guesses that pass a quick review but miss real edge cases, such as unhandled retry conditions or stale references after cleanup. The fix is writing exact behavior, non-goals, failure modes, and test cases before requesting code, turning vague thinking into something safe to execute rather than discovering the gaps in production. _Engineers refining how they scope AI coding tasks can find more workflow write-ups like this on daily.dev._ ### Why shouldn't I ask the same AI chat session that wrote code to also review it? A session that produced a diff already carries the assumptions behind it, so it tends to catch surface issues like missing tests or naming inconsistencies but rarely challenges its own design decisions. Running the review in a separate context, or with a different tool, using explicit prompts for race conditions, unrelated changes, and unnecessary fixtures produces more genuine risk discovery. _Teams building review habits around AI-generated code can compare approaches like this via daily.dev._ ### How should I manage context when working across multiple AI coding sessions on the same codebase? Treat context as a finite resource rather than a landfill: start a new conversation for each new bug, keep refactor discussions separate from implementation, and separate the review phase from planning. Persisting important decisions in a task note or specification instead of a chat transcript prevents the AI from confidently connecting unrelated things pasted into one long session. _Developers structuring long AI pairing sessions often track workflow lessons like this on daily.dev._ ## Community discussion Top comments from developers on daily.dev. **@bullit2k** · 1 upvotes > Enjoyable article. Keep AI on a short leash and feed it good spec. I like the idea of keeping diff size down. Lots of good tips in here to strengthen my workflow and to ensure the end result is my responsibility and fully owned and understood. Thanks Mohamed. **@kartiknvj** · 1 upvotes > The lesson that lands hardest for me is never letting the same model review its own output, because a model scoring its own work carries a self-preference bias and will wave through the thing it just wrote. Spec-driven development framed as good engineering that outlasts the tool is the right way to hold it, since small diffs and clear requirements were correct long before agents. Treating context as a finite budget is underrated too, most of the wrong outputs I get trace back to a task that was too big for one pass. **@erikandersoon** · 0 upvotes > The author's framing that AI rewards good engineering is the most optimistic take on this shift. It's tempting to think AI makes sloppy work faster, but the reality is that AI amplifies whatever habits you already have. If you have weak specs, AI gives you weak code faster. If you skip testing, AI gives you untested code faster. But if you practice spec-driven development, small changes, and independent review, AI becomes a force multiplier for good engineering. The habits that AI rewards are the habits that make you a better engineer regardless of the tool. ## 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 - [Engineering Large and Complex Code Using AI: A Practitioner’s View](https://daily.dev/posts/engineering-large-and-complex-code-using-ai-a-practitioner-s-view-zr49vlfr2) · Medium · 1 upvotes · 0 comments - [On AI](https://daily.dev/posts/on-ai-y4nttoreo) · openmymind · 0 upvotes · 0 comments - [From Implementers to Solution Architects](https://daily.dev/posts/from-implementers-to-solution-architects-kqx8z1vba) · agoda · 0 upvotes · 0 comments - [AI still needs humans](https://daily.dev/posts/ai-still-needs-humans-nml187ik0) · InfoWorld · 0 upvotes · 0 comments --- Tags: [#testing](https://daily.dev/tags/testing), [#ai-coding](https://daily.dev/tags/ai-coding), [#code-review](https://daily.dev/tags/code-review), [#spec-driven-development](https://daily.dev/tags/spec-driven-development) [View this post on daily.dev](https://daily.dev/posts/what-ai-secretly-changed-about-my-engineering-71x5m4e3y) ```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":"What AI Secretly Changed About My Engineering","url":"https://daily.dev/posts/what-ai-secretly-changed-about-my-engineering-71x5m4e3y","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/what-ai-secretly-changed-about-my-engineering-71x5m4e3y"},"datePublished":"2026-07-08T07:05:32.358Z","dateModified":"2026-09-14T09:07:59.317Z","description":"After hundreds of hours using AI coding tools like Claude, Copilot, and JetBrains Junie on real production services, the author found that AI didn't just make...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/f7195056be7140138e9d9f4004f68bc6?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/f7195056be7140138e9d9f4004f68bc6?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"iO tech_hub","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":"iO tech_hub","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/ac540f589d8843d99692bd6368ee7680","url":"https://daily.dev/sources/iotechhub"},"commentCount":3,"discussionUrl":"https://daily.dev/posts/what-ai-secretly-changed-about-my-engineering-71x5m4e3y","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":27},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":3}],"keywords":"testing,ai-coding,code-review,spec-driven-development","timeRequired":"PT15M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"iO tech_hub","item":"https://daily.dev/sources/iotechhub"},{"@type":"ListItem","position":3,"name":"What AI Secretly Changed About My Engineering"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/what-ai-secretly-changed-about-my-engineering-71x5m4e3y","comment":[{"@type":"Comment","text":"Enjoyable article. Keep AI on a short leash and feed it good spec. I like the idea of keeping diff size down. Lots of good tips in here to strengthen my workflow and to ensure the end result is my responsibility and fully owned and understood. Thanks Mohamed.","datePublished":"2026-07-08T12:42:06.387Z","dateModified":"2026-07-08T12:43:36.029Z","url":"https://daily.dev/posts/71X5M4E3Y#c-Sfqg0WK9z","author":{"@type":"Person","name":"Gordon","url":"https://daily.dev/bullit2k","image":"https://lh3.googleusercontent.com/a-/AOh14GiKnakwOvfjp6wNSamn-lCg3wUp8j_5LVLPZNkpfQ=s100"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1}},{"@type":"Comment","text":"The lesson that lands hardest for me is never letting the same model review its own output, because a model scoring its own work carries a self-preference bias and will wave through the thing it just wrote. Spec-driven development framed as good engineering that outlasts the tool is the right way to hold it, since small diffs and clear requirements were correct long before agents. Treating context as a finite budget is underrated too, most of the wrong outputs I get trace back to a task that was too big for one pass.","datePublished":"2026-07-09T17:24:33.711Z","url":"https://daily.dev/posts/71X5M4E3Y#c-y66lgkZWt","author":{"@type":"Person","name":"kartik-nvjk","url":"https://daily.dev/kartiknvj","image":"https://media.daily.dev/image/upload/s--3gGgsVCw--/f_auto/v1781456774/avatars/avatar_TvTVeiMdkRCqWUDullFmy?_a=BAMAMiWQ0"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1}},{"@type":"Comment","text":"The author’s framing that AI rewards good engineering is the most optimistic take on this shift. It’s tempting to think AI makes sloppy work faster, but the reality is that AI amplifies whatever habits you already have. If you have weak specs, AI gives you weak code faster. If you skip testing, AI gives you untested code faster. But if you practice spec-driven development, small changes, and independent review, AI becomes a force multiplier for good engineering. The habits that AI rewards are the habits that make you a better engineer regardless of the tool.","datePublished":"2026-07-11T23:02:41.889Z","url":"https://daily.dev/posts/71X5M4E3Y#c-mWA98Z3Ny","author":{"@type":"Person","name":"Erik Andersoon","url":"https://daily.dev/erikandersoon","image":"https://lh3.googleusercontent.com/a/ACg8ocIjscYXslkLRHxEihtWWVgTFELNg4TVNJzpD0qugegOqLsZAw=s96-c"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/what-ai-secretly-changed-about-my-engineering-71x5m4e3y#faq","mainEntity":[{"@type":"Question","name":"Why does using AI coding assistants like Claude or Copilot make ambiguous requirements more costly?","acceptedAnswer":{"@type":"Answer","text":"Vague task descriptions let AI fill gaps with plausible-sounding guesses that pass a quick review but miss real edge cases, such as unhandled retry conditions or stale references after cleanup. The fix is writing exact behavior, non-goals, failure modes, and test cases before requesting code, turning vague thinking into something safe to execute rather than discovering the gaps in production. Engineers refining how they scope AI coding tasks can find more workflow write-ups like this on daily.dev."}},{"@type":"Question","name":"Why shouldn't I ask the same AI chat session that wrote code to also review it?","acceptedAnswer":{"@type":"Answer","text":"A session that produced a diff already carries the assumptions behind it, so it tends to catch surface issues like missing tests or naming inconsistencies but rarely challenges its own design decisions. Running the review in a separate context, or with a different tool, using explicit prompts for race conditions, unrelated changes, and unnecessary fixtures produces more genuine risk discovery. Teams building review habits around AI-generated code can compare approaches like this via daily.dev."}},{"@type":"Question","name":"How should I manage context when working across multiple AI coding sessions on the same codebase?","acceptedAnswer":{"@type":"Answer","text":"Treat context as a finite resource rather than a landfill: start a new conversation for each new bug, keep refactor discussions separate from implementation, and separate the review phase from planning. Persisting important decisions in a task note or specification instead of a chat transcript prevents the AI from confidently connecting unrelated things pasted into one long session. Developers structuring long AI pairing sessions often track workflow lessons like this on daily.dev."}}]} ```