--- title: Latent Space's Frontier AEO Tracker: what frontier LLMs... description: Latent Space launched a Frontier AEO Tracker built on its Astra research pipeline, measuring what recommendations frontier LLMs (Claude Opus/Fable, GPT... canonical: https://daily.dev/posts/latent-space-s-frontier-aeo-tracker-what-frontier-llms-recommend-and-why-it-matters-mdtyofkka twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Latent Space's Frontier AEO Tracker: what frontier LLMs recommend and why it matters | daily.dev og:description: Latent Space launched a Frontier AEO Tracker built on its Astra research pipeline, measuring what recommendations frontier LLMs (Claude Opus/Fable, GPT... og:url: https://daily.dev/posts/latent-space-s-frontier-aeo-tracker-what-frontier-llms-recommend-and-why-it-matters-mdtyofkka og:image: https://api.daily.dev/og/posts/mDtYOFKka.png og:image:alt: Latent Space's Frontier AEO Tracker: what frontier LLMs recommend and why it matters 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. # Latent Space's Frontier AEO Tracker: what frontier LLMs recommend and why it matters **[Collections](https://daily.dev/sources/collections)** · 3 min read · 1 upvotes · 0 comments ## Summary Latent Space launched a Frontier AEO Tracker built on its Astra research pipeline, measuring what recommendations frontier LLMs (Claude Opus/Fable, GPT Sol/Astra, Grok, Muse, SWE) give across 161 categories like coding agents, AI podcasts, and databases. Findings include strong self-bias (OpenAI models push Codex, Grok pushes Cursor), 28 categories with a dominant consensus pick, meaningful behavioral differences across model generations in retrieval strategy, and validation that AEO tactics like markdown content-negotiation genuinely affect whether a model reads a site at all. The team frames this as an early attempt to make answer engine optimization legible the way SEO tooling once did for search. ## Content ## What the Frontier AEO Tracker found Latent Space recently launched a Frontier AEO (Answer Engine Optimization) Tracker built on their Astra research pipeline. The project analyzes what recommendations frontier LLMs give across 161 categories - coding agents, AI podcasts, databases, and more. It runs 6 prompt variations across 7 models (Claude Opus/Fable, GPT Sol/Astra, Grok, Muse, SWE) and scores results with a proprietary AEO score that weights first choices, alternatives, mentions, and anti-recommendations. A few things stood out: **Self-bias is real.** Models tend to favor their own lab's products. OpenAI's models push Codex; Grok recommends Cursor. This isn't subtle. **28 categories have a dominant consensus pick** - one tool that basically every model agrees on regardless of prompt variation. **Model generations behave differently.** Astra (GPT's newer version) searches fewer sources than Sol but gives more confident, consistent answers. Fable searches more sources than Opus. Whether that's better or worse depends on what you're optimizing for. The team also validated something practically useful: AEO practices like markdown content-negotiation genuinely affect whether models actually read a site's content when forming recommendations. ## What about llms.txt? There's been a lot of noise about llms.txt - the proposed Markdown file you place at a site's root to help AI models understand what you do. The pitch is that it's like robots.txt but for LLMs. The honest answer in 2026: it probably doesn't move the needle on citations or rankings. A 100,000-site study found only correlation, not causation. Technically mature sites tend to do many things well at once - clean structure, good documentation, third-party mentions - and publishing llms.txt is just one of those things. Major answer engines rarely request the file directly, and most of the requests that do show up in server logs come from SEO tools checking whether the file exists, not actual crawlers. That said, it's not useless. For sites with developer docs or complex APIs, llms.txt can serve as reasonable documentation hygiene - it helps AI coding agents navigate your content once they've already arrived. It's just not a ranking signal. ## What actually drives AI citations Based on both the Frontier AEO Tracker findings and the llms.txt research, a clearer picture emerges: - **Clarity about what your product does.** Models can't recommend something they can't categorize. - **Independent corroboration.** Reviews, third-party mentions, and discussions across multiple sources matter more than anything on your own site. - **Pages that resolve real decisions.** Useful comparisons, honest tradeoffs, and specific use cases outperform marketing copy. - **Technical accessibility.** Markdown content-negotiation and clean structure affect whether models read your content at all. The file at your root matters less than whether the broader web has a coherent, consistent story about what you do and why someone should use it. ## Questions this post answers ### Does adding an llms.txt file to my site improve how often AI models recommend it? Not meaningfully as a ranking signal. A 100,000-site study found only correlation, not causation, between having llms.txt and better AI visibility, since technically mature sites tend to already do many things well. Major answer engines rarely request the file directly, and most server-log requests come from SEO tools checking existence rather than real crawlers. It still helps as documentation hygiene for AI coding agents once they arrive. _Developers deciding whether llms.txt is worth their time can track this AEO research on daily.dev._ ### Why do AI models like GPT and Grok recommend their own company's products more often? Self-bias shows up clearly in frontier model recommendations: OpenAI's models push Codex while Grok recommends Cursor rather than competing products from other labs. Testing across 161 categories with 6 prompt variations on 7 models (Claude Opus/Fable, GPT Sol/Astra, Grok, Muse, SWE) found this bias consistent and not subtle, though 28 categories still reached a dominant cross-model consensus pick. _Anyone comparing AI coding tools can follow self-bias findings like this on daily.dev before trusting a chatbot's pick._ ## Similar posts on daily.dev - [LLMs.txt Explained: The Ultimate 2026 Guide to AI Search, GEO, AI Crawlers, and LLM Optimization](https://daily.dev/posts/llms-txt-explained-the-ultimate-2026-guide-to-ai-search-geo-ai-crawlers-and-llm-optimization-dh1nqdngp) · C\# Corner · 0 upvotes · 0 comments - [How brands can influence visibility in AI-generated answers](https://daily.dev/posts/how-brands-can-influence-visibility-in-ai-generated-answers-mjoxamr4l) · The Next Web · 0 upvotes · 0 comments - [How to Measure Your AI Citation Rate Across ChatGPT, Perplexity, and Claude](https://daily.dev/posts/how-to-measure-your-ai-citation-rate-across-chatgpt-perplexity-and-claude-hsdkfvagk) · freeCodeCamp · 4 upvotes · 2 comments - [Generative Engine Optimization \(GEO\) Guide for 2026](https://daily.dev/posts/generative-engine-optimization-geo-guide-for-2026-gayl8hwhr) · Alex CloudStar · 0 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#seo](https://daily.dev/tags/seo) [View this post on daily.dev](https://daily.dev/posts/latent-space-s-frontier-aeo-tracker-what-frontier-llms-recommend-and-why-it-matters-mdtyofkka) ```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":"Latent Space's Frontier AEO Tracker: what frontier LLMs recommend and why it matters","url":"https://daily.dev/posts/latent-space-s-frontier-aeo-tracker-what-frontier-llms-recommend-and-why-it-matters-mdtyofkka","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/latent-space-s-frontier-aeo-tracker-what-frontier-llms-recommend-and-why-it-matters-mdtyofkka"},"datePublished":"2026-09-07T22:00:46.351Z","dateModified":"2026-09-13T20:06:00.003Z","description":"Latent Space launched a Frontier AEO Tracker built on its Astra research pipeline, measuring what recommendations frontier LLMs (Claude Opus/Fable, GPT...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1524de7194e8a94423b473fd0094f184?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1524de7194e8a94423b473fd0094f184?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Collections","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":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/latent-space-s-frontier-aeo-tracker-what-frontier-llms-recommend-and-why-it-matters-mdtyofkka","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,seo","timeRequired":"PT3M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"Latent Space's Frontier AEO Tracker: what frontier LLMs recommend and why it matters"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/latent-space-s-frontier-aeo-tracker-what-frontier-llms-recommend-and-why-it-matters-mdtyofkka#faq","mainEntity":[{"@type":"Question","name":"Does adding an llms.txt file to my site improve how often AI models recommend it?","acceptedAnswer":{"@type":"Answer","text":"Not meaningfully as a ranking signal. A 100,000-site study found only correlation, not causation, between having llms.txt and better AI visibility, since technically mature sites tend to already do many things well. Major answer engines rarely request the file directly, and most server-log requests come from SEO tools checking existence rather than real crawlers. It still helps as documentation hygiene for AI coding agents once they arrive. Developers deciding whether llms.txt is worth their time can track this AEO research on daily.dev."}},{"@type":"Question","name":"Why do AI models like GPT and Grok recommend their own company's products more often?","acceptedAnswer":{"@type":"Answer","text":"Self-bias shows up clearly in frontier model recommendations: OpenAI's models push Codex while Grok recommends Cursor rather than competing products from other labs. Testing across 161 categories with 6 prompt variations on 7 models (Claude Opus/Fable, GPT Sol/Astra, Grok, Muse, SWE) found this bias consistent and not subtle, though 28 categories still reached a dominant cross-model consensus pick. Anyone comparing AI coding tools can follow self-bias findings like this on daily.dev before trusting a chatbot's pick."}}]} ```