--- title: Make no assumptions. | daily.dev description: A reflection on how AI coding harnesses and LLMs are accelerating the problem of 'messy reasoning horizons' — analogous to layered soil horizons in codebases —... canonical: https://daily.dev/posts/make-no-assumptions--5b9ipyvvt twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Make no assumptions. | daily.dev og:description: A reflection on how AI coding harnesses and LLMs are accelerating the problem of 'messy reasoning horizons' — analogous to layered soil horizons in codebases —... og:url: https://daily.dev/posts/make-no-assumptions--5b9ipyvvt og:image: https://api.daily.dev/og/posts/5B9IPyVVT.png og:image:alt: Make no assumptions. 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. # Make no assumptions. **[Irrational Exuberance](https://daily.dev/sources/lethain)** · 9 min read · 3 upvotes · 0 comments ## Summary A reflection on how AI coding harnesses and LLMs are accelerating the problem of 'messy reasoning horizons' — analogous to layered soil horizons in codebases — where flawed reasoning layers compound into catastrophic decisions. The author shares a concrete example of an incident analysis that nearly led to solving a non-existent problem due to unvalidated LLM output. Five practical approaches are proposed: making no assumptions about any reasoning context, ensuring authors are the first human reviewer of their own output, prioritizing deterministic software conventions, separating experimentation from production work, and structuring LLM use to exploration and refinement rather than direct conclusions. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How can relying on LLMs for reasoning lead to wrong business conclusions? Trusting LLM output without inspection can poison an entire reasoning layer, since once a flawed conclusion is accepted, everything built on top of it inherits the error while still looking like solid strategic work. An engineering leader described pulling incident data via an MCP that appeared to show fewer data-related incidents than expected, only to discover the incident tool omitted roughly half of relevant incidents that weren't properly tagged with a team. Accepting the first analysis would have led to solving an imaginary problem. _Developers wary of trusting AI-generated analysis without verification can find similar cautionary experiences on daily.dev._ ### What practices help prevent flawed reasoning when using LLMs and coding harnesses at work? Key practices include making no assumptions and inspecting every piece of information before accepting it, ensuring the content's creator is always the first human reviewer before passing work to others, prioritizing deterministic conventions over ambiguous ones since harnesses don't handle implicit norms well, and separating low-stakes play from high-stakes production work so mistakes happen while learning rather than in critical decisions. _Teams building AI-assisted workflows can track emerging best practices for reasoning with LLMs on daily.dev._ ## Similar posts on daily.dev - [Stop generating, start thinking](https://daily.dev/posts/stop-generating-start-thinking-7jjvm8uwh) · Hacker News · 16 upvotes · 1 comments - [How to Work with AI Coding Tools Without Losing Your Engineering Judgment](https://daily.dev/posts/how-to-work-with-ai-coding-tools-without-losing-your-engineering-judgment-qnygkubp4) · IT Revolution · 1 upvotes · 0 comments - [How to use LLMs without driving your team mad](https://daily.dev/posts/how-to-use-llms-without-driving-your-team-mad-bcztby6jm) · thoughbot · 0 upvotes · 0 comments - [Fragments: February 9](https://daily.dev/posts/fragments-february-9-doxsl4jom) · Martin Fowler · 1 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#architecture](https://daily.dev/tags/architecture), [#leadership](https://daily.dev/tags/leadership), [#ai-coding](https://daily.dev/tags/ai-coding), [#technical-debt](https://daily.dev/tags/technical-debt) [View this post on daily.dev](https://daily.dev/posts/make-no-assumptions--5b9ipyvvt) ```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":"Make no assumptions.","url":"https://daily.dev/posts/make-no-assumptions--5b9ipyvvt","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/make-no-assumptions--5b9ipyvvt"},"datePublished":"2026-07-11T15:16:55.520Z","dateModified":"2026-09-14T06:11:40.542Z","description":"A reflection on how AI coding harnesses and LLMs are accelerating the problem of 'messy reasoning horizons' — analogous to layered soil horizons in codebases —...","image":"https://media.daily.dev/image/upload/s--0_ODbtD2--/f_auto/v1722860399/public/Placeholder%2008","thumbnailUrl":"https://media.daily.dev/image/upload/s--0_ODbtD2--/f_auto/v1722860399/public/Placeholder%2008","isAccessibleForFree":true,"articleSection":"Irrational Exuberance","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":"Irrational Exuberance","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/e6ab29f02f344049b087fabd480dcaf0","url":"https://daily.dev/sources/lethain"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/make-no-assumptions--5b9ipyvvt","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":3},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,architecture,leadership,ai-coding,technical-debt","timeRequired":"PT9M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Irrational Exuberance","item":"https://daily.dev/sources/lethain"},{"@type":"ListItem","position":3,"name":"Make no assumptions."}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/make-no-assumptions--5b9ipyvvt#faq","mainEntity":[{"@type":"Question","name":"How can relying on LLMs for reasoning lead to wrong business conclusions?","acceptedAnswer":{"@type":"Answer","text":"Trusting LLM output without inspection can poison an entire reasoning layer, since once a flawed conclusion is accepted, everything built on top of it inherits the error while still looking like solid strategic work. An engineering leader described pulling incident data via an MCP that appeared to show fewer data-related incidents than expected, only to discover the incident tool omitted roughly half of relevant incidents that weren't properly tagged with a team. Accepting the first analysis would have led to solving an imaginary problem. Developers wary of trusting AI-generated analysis without verification can find similar cautionary experiences on daily.dev."}},{"@type":"Question","name":"What practices help prevent flawed reasoning when using LLMs and coding harnesses at work?","acceptedAnswer":{"@type":"Answer","text":"Key practices include making no assumptions and inspecting every piece of information before accepting it, ensuring the content's creator is always the first human reviewer before passing work to others, prioritizing deterministic conventions over ambiguous ones since harnesses don't handle implicit norms well, and separating low-stakes play from high-stakes production work so mistakes happen while learning rather than in critical decisions. Teams building AI-assisted workflows can track emerging best practices for reasoning with LLMs on daily.dev."}}]} ```