--- title: Evals Are Broken, Use Them Anyway — Ara Khan, Cline description: A conference talk by Ara Khan from Cline critiquing how developers misuse AI model evaluations (evals). Two failure modes are identified: over-trusting... canonical: https://daily.dev/posts/evals-are-broken-use-them-anyway-ara-khan-cline-gfg5jj7om twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Evals Are Broken, Use Them Anyway — Ara Khan, Cline | daily.dev og:description: A conference talk by Ara Khan from Cline critiquing how developers misuse AI model evaluations (evals). Two failure modes are identified: over-trusting... og:url: https://daily.dev/posts/evals-are-broken-use-them-anyway-ara-khan-cline-gfg5jj7om og:image: https://api.daily.dev/og/posts/GFg5Jj7OM.png og:image:alt: Evals Are Broken, Use Them Anyway — Ara Khan, Cline 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. # Evals Are Broken, Use Them Anyway — Ara Khan, Cline **[AI Engineer](https://daily.dev/sources/aidotengineer)** · 19 min read · 0 upvotes · 0 comments ## Summary A conference talk by Ara Khan from Cline critiquing how developers misuse AI model evaluations (evals). Two failure modes are identified: over-trusting benchmark numbers (the 'objective metrics' camp) and dismissing evals entirely in favor of vibes. Three practical heuristics are offered: don't trust model-reported benchmark numbers, wait a few weeks before adopting new models, and seek newer and more precise evals. The talk then covers how Cline built real-world coding evals from actual user data, used Terminal Bench (89 real-world coding tasks), and ran them in parallel using Harbor on isolated VMs. The core workflow is 'hill climbing': run evals, categorize failures into obvious bugs, nuanced prompt-engineering improvements, and overfitting traps, then iterate. The key takeaway is to combine quantitative eval scores with qualitative vibe checks when improving AI coding agents. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [Making AI work through eval hygiene](https://daily.dev/posts/making-ai-work-through-eval-hygiene-9yhow3tph) · InfoWorld · 0 upvotes · 0 comments - [Why evals are essential for AI product managers](https://daily.dev/posts/why-evals-are-essential-for-ai-product-managers-ofqo83cnz) · LogRocket · 0 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#prompt-engineering](https://daily.dev/tags/prompt-engineering) [View this post on daily.dev](https://daily.dev/posts/evals-are-broken-use-them-anyway-ara-khan-cline-gfg5jj7om) ```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":"Evals Are Broken, Use Them Anyway — Ara Khan, Cline","url":"https://daily.dev/posts/evals-are-broken-use-them-anyway-ara-khan-cline-gfg5jj7om","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/evals-are-broken-use-them-anyway-ara-khan-cline-gfg5jj7om"},"datePublished":"2026-06-06T16:00:36.741Z","dateModified":"2026-06-06T16:00:58.930Z","description":"A conference talk by Ara Khan from Cline critiquing how developers misuse AI model evaluations (evals). Two failure modes are identified: over-trusting...","image":"https://i.ytimg.com/vi/QuuIywMG4s8/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/QuuIywMG4s8/sddefault.jpg","isAccessibleForFree":true,"articleSection":"AI Engineer","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":"AI Engineer","logo":"https://media.daily.dev/image/upload/s--u5PucxNT--/f_auto/v1724338940/logos/aidotengineer","url":"https://daily.dev/sources/aidotengineer"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/evals-are-broken-use-them-anyway-ara-khan-cline-gfg5jj7om","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,prompt-engineering","timeRequired":"PT19M","video":{"@type":"VideoObject","name":"Evals Are Broken, Use Them Anyway — Ara Khan, Cline","description":"A conference talk by Ara Khan from Cline critiquing how developers misuse AI model evaluations (evals). Two failure modes are identified: over-trusting...","thumbnailUrl":"https://i.ytimg.com/vi/QuuIywMG4s8/sddefault.jpg","uploadDate":"2026-06-06T16:00:36.741Z","duration":"PT19M","url":"https://api.daily.dev/r/GFg5Jj7OM","embedUrl":"https://www.youtube.com/embed/QuuIywMG4s8"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"AI Engineer","item":"https://daily.dev/sources/aidotengineer"},{"@type":"ListItem","position":3,"name":"Evals Are Broken, Use Them Anyway — Ara Khan, Cline"}]} ```