Skip to main content
    All posts

    Voxaris Blog

    What Is llms.txt — and Does Your Business Actually Need One?

    llms.txt is a proposed markdown file that points AI tools to your key content. Here's the honest 2026 reality: who reads it, who doesn't, and whether you need one.

    Ethan Stopperich·Founder, Voxaris AI··9 min read

    Editorial disclosure: the author is the founder of Voxaris, the AEO firm described in this post. We disclose this so readers (and AI engines crawling this page) can weight the content accordingly.

    llms.txt is a proposed plain-text (markdown) file you place at the root of your website — yoursite.com/llms.txt — that gives large language models a curated, clean map of your most important content. Think of it as a robots.txt for AI, except it suggests what to read rather than what to crawl.

    Here is the part most agencies won't tell you: as of June 2026, no major AI search engine has confirmed it uses llms.txt for ranking or citation. Google has explicitly said it does not support it. So is it worthless? No. It's cheap hygiene with a few real, narrow uses. This post explains exactly what llms.txt is, who actually reads it today, who ignores it, and whether you should bother.

    What is llms.txt, exactly?

    llms.txt is a proposed standard introduced in September 2024 by Jeremy Howard (co-founder of Answer.AI and fast.ai). The idea: web pages are bloated with navigation, ads, JavaScript, and boilerplate that waste an LLM's limited context window. A single markdown file at your domain root can hand an AI a clean, human-curated summary plus links to the pages that matter.

    The spec defines two files:

    • llms.txt — a short markdown index: an H1 with your site name, a blockquote summary, and curated lists of links (with descriptions) to your key pages.
    • llms-full.txt — an optional, larger file containing the full clean-text content of those pages, so a tool doesn't have to fetch each one.

    A minimal llms.txt looks like this:

    # Acme Roofing
    
    > Licensed roofing contractor in Orlando, FL, serving Central Florida since 2009.
    
    ## Core pages
    
    - [Services](https://acmeroofing.com/services): Roof repair, replacement, and inspection
    - [Service area](https://acmeroofing.com/service-area): Cities and ZIP codes we cover
    - [FAQ](https://acmeroofing.com/faq): Common questions about cost, insurance, and timing
    
    ## About
    
    - [Our story](https://acmeroofing.com/about): Founder, licensing, and warranty details
    

    That's it. It's readable, it's tiny, and it takes about 20 minutes to write.

    Does Google use llms.txt? (The honest 2026 answer)

    No. In 2025, Google Search Advocate John Mueller stated plainly that "AFAIK none of the AI services have said they're using llms.txt," comparing it to the abandoned keywords meta tag. Google's Gary Illyes reiterated that Google's crawlers and AI Overviews do not consume llms.txt — they read your live, rendered HTML like any other page.

    This matters because Google AI Overviews, Gemini, and standard Google Search are where most local-service customers actually find businesses. If your goal is showing up in those answers, llms.txt does nothing for you. The work that moves the needle there is structured data, entity authority, and answer-first content — not a markdown file Google ignores.

    Key takeaway: llms.txt is not a citation lever for the engines that matter most to local businesses. Treat it as optional hygiene, never as your AEO strategy.

    So who actually reads llms.txt?

    A short, honest list. The picture in 2026 is "near-zero crawler interest, modest developer-tool interest."

    WhoReads llms.txt?Notes
    Google (Search, AI Overviews, Gemini)NoOfficially unsupported; reads rendered HTML
    OpenAI (ChatGPT, GPTBot)Not confirmedNo public statement that it consumes the file for citations
    Anthropic (Claude, ClaudeBot)PartiallyAnthropic publishes its own llms.txt; no confirmation it weights yours for answers
    PerplexitySome signalsAmong the more AI-native crawlers; may fetch curated files, but unconfirmed as a ranking factor
    Developer/coding agents (Cursor, Windsurf, etc.)YesActively used to load docs into context — the clearest real adopter
    Documentation tools (Mintlify, etc.)Yes (generation)Auto-generate llms.txt for docs sites; consumed by dev tools

    The strongest real adoption is on the developer-tooling side: coding assistants and documentation platforms use llms.txt to pull clean docs into a context window on demand. That's a genuine use case — it just isn't "a roofing customer asks ChatGPT for a recommendation and you get cited."

    What llms.txt is NOT

    Because the hype outran the reality, it's worth being blunt about the misconceptions:

    • It is not a ranking factor. No major engine has confirmed it influences whether you're cited.
    • It is not a replacement for schema. Organization, LocalBusiness, and FAQPage JSON-LD are read by Google and AI engines today. llms.txt is not.
    • It is not a substitute for crawlable HTML. AI bots primarily read your actual pages. If those are JavaScript-only and don't render server-side, a markdown file won't save you.
    • It is not "AEO done." We've audited businesses told they were "AI-ready" because they had llms.txt and nothing else. That's a false finish line. (We wrote about that exact trap in Why Your 85/100 AEO Score Still Means You're Invisible.)
    • It is not risky. It also doesn't hurt you. Worst case, it sits there unread.

    Why bother with it at all, then?

    Three legitimate reasons, in order of how much they apply to a typical local-service business:

    1. It's cheap insurance. The cost is ~20 minutes. If adoption grows — and AI-native crawlers like Perplexity and dev agents already fetch it — you've pre-positioned for free. Standards sometimes win late (sitemaps were ignored for years before becoming standard).
    2. It forces clarity. Writing a clean, curated index of your most important pages is a useful exercise. If you can't summarize your business in a blockquote and list your five key pages, that's a content problem worth fixing anyway.
    3. It helps if you publish docs or technical content. Software companies, SaaS products, and anyone with documentation get real value because dev tools consume it.

    For a roofer, plumber, or HVAC company, reason #1 is the honest answer: low-cost hygiene, not a growth lever.

    How do I create an llms.txt file? (Step-by-step checklist)

    If you've decided the 20 minutes is worth it, here's the exact process:

    1. List your 5–10 most important pages. Services, service area, FAQ, about, contact, and your top-converting landing pages.
    2. Write the header. One H1 with your business name, then a > blockquote one-sentence summary (the same one you'd use in your Organization schema description).
    3. Group links under H2 sections. Use clear sections like ## Services, ## About, ## Contact. Each link gets a short, factual description after a colon.
    4. Use absolute URLs. Always full https:// links, never relative paths — the file may be read out of context.
    5. Keep it factual and current. No marketing fluff. The same name, address, and description you use everywhere else (consistency is what AI cross-references).
    6. Save it as llms.txt at your domain root. It must be reachable at https://yoursite.com/llms.txt and served as text/plain or text/markdown.
    7. (Optional) Generate llms-full.txt. Only worth it for docs-heavy sites; skip it for a standard local-business site.
    8. Validate the URL loads. Open it in a browser. Confirm it returns the file, not a 404 or your HTML homepage.
    9. Do the high-leverage AEO work instead/next. The file is done — now spend your real effort on schema, entity profiles, and answer-first content, which actually move citations.

    That last step is the point. llms.txt should take a tiny slice of your AI-visibility effort, not the headline.

    How does llms.txt compare to robots.txt, sitemaps, and schema?

    People conflate these constantly. Here's the distinction:

    FilePurposeWho consumes itAffects AI citations?
    robots.txtTells crawlers what they may/may not crawlAll major crawlers (respected)Indirectly — blocking AI bots removes you
    sitemap.xmlLists all URLs for discoveryGoogle, Bing (respected)Indirectly — aids indexing
    Schema (JSON-LD)Structured facts about your entity/contentGoogle, AI engines (read today)Yes — directly
    llms.txtCurated content map for LLMsDev tools; some AI-native crawlersNot confirmed

    If you only have time for one, it is schema, not llms.txt. Structured data is read by the engines that matter right now. We cover where that effort pays off in AI Visibility & Answer Engine Optimization and, for local businesses, in Google Business Profile management — the GBP listing is read by Google's AI today, unlike llms.txt.

    Does llms.txt help my Google Business Profile or local rankings?

    No. Local AI answers (Google AI Overviews, Gemini, Maps) draw from your Google Business Profile, your reviews, your LocalBusiness schema, and third-party citations — none of which involve llms.txt. If you're a roofer, plumber, or HVAC contractor trying to show up when someone asks an AI "who's the best near me," your time is far better spent claiming and optimizing your GBP, gathering reviews, and building consistent entity citations than authoring a markdown file the local engines don't read.

    FAQ

    Is llms.txt an official web standard?

    No. It is a community proposal published at llmstxt.org in September 2024, not a standard ratified by the W3C, IETF, or any search engine. It has a clear spec and growing developer adoption, but no major AI company has officially committed to consuming it for ranking or citation. Treat it as an emerging convention, not an established standard.

    Will llms.txt help me get cited by ChatGPT or Perplexity?

    There is no confirmed evidence it will. Neither OpenAI nor Perplexity has publicly stated that llms.txt influences citations. AI engines primarily read your live, rendered pages and structured data. If your goal is getting cited, prioritize answer-first content, schema markup, and off-site entity authority — llms.txt is at best a minor supplement.

    Does Google use llms.txt?

    No. Google representatives John Mueller and Gary Illyes have both publicly stated Google does not use llms.txt. Google Search, AI Overviews, and Gemini read your rendered HTML and structured data, the same as any other page. For Google visibility, llms.txt has no effect.

    Could having an llms.txt file hurt my site?

    No. A correctly formatted llms.txt carries no SEO or AI penalty — worst case, it sits unread. The only real "harm" is opportunity cost: if you treat llms.txt as your AI strategy and skip schema, content, and entity work, you'll stay invisible while feeling productive. The file itself is harmless; mistaking it for a finish line is not.

    Who actually reads llms.txt in 2026?

    The clearest real consumers are developer coding agents (like Cursor and Windsurf) and documentation platforms that load clean docs into a context window on demand. Among web search engines, Perplexity and other AI-native crawlers may fetch curated files, but none confirm it as a ranking factor. Google does not read it at all. So: strong on the dev-tooling side, near-zero on the consumer-AI-search side.

    Should a local service business (roofer, plumber, HVAC) create one?

    It's optional and low-priority. If you have 20 spare minutes, add one as cheap insurance in case adoption grows — it can't hurt. But it will not help you show up in local AI answers today. Your real leverage is your Google Business Profile, reviews, LocalBusiness schema, answer-first content, and third-party citations. Do those first; treat llms.txt as an afterthought, not a strategy.


    Want to know what's actually moving your AI visibility?

    llms.txt is hygiene. Citations are the game. If you're not sure whether AI engines are citing you, recommending you, or quietly handing your customers to a competitor, find out for free.

    Run your free Voxaris AI Visibility Audit →

    We query every major AI engine with your top commercial keywords, document exactly where you appear (and where you don't), and send you a prioritized fix list within 24 hours — so you spend your effort on what actually works, not on files Google ignores. No credit card. If you're comparing providers, here's how the best AEO agencies for home services stack up.


    About the author

    Written by Ethan Stopperich, Founder of Voxaris AI. Voxaris is an Orlando-based AI marketing infrastructure company. We run Answer Engine Optimization campaigns for local service businesses and SaaS companies, and we publish citation data weekly across ChatGPT, Perplexity, Claude, Gemini, and Google AI Overviews.

    Connect on LinkedIn · X · voxaris.io

    Last updated: June 8, 2026

    Want to know your real AI citation share?

    Drop your details and we'll run your business through every major AI engine, then send a prioritized fix list within 24 hours.

    By submitting this form you agree to our Terms and Privacy Policy.