llms.txt is a plain-text Markdown file placed at your website's root (yourdomain.com/llms.txt) that tells AI models what your site is about, which pages matter most, and where to find them. Think of it as a sitemap built for AI instead of search engines. The format follows the llmstxt.org specification proposed by Jeremy Howard. With only about 10% of websites having one (according to ALLMO), early adopters have a clear differentiation opportunity.
- What it is: A plain-text Markdown file at your site's root (/llms.txt)
- Purpose: Gives AI crawlers a structured summary of your most important content
- Format: H1 heading, blockquote description, H2 sections with link lists
- Adoption: Only ~10% of websites have one - early-mover advantage is significant
- Effort: 15 minutes to create with our free generator, zero downside
How Does llms.txt Work?
When an AI model visits your site, it has limited context. It cannot read every page the way a human browses. llms.txt gives the model a curated summary: your company name, a one-line description, and links to your most important pages with brief explanations of what each one contains. According to Mintlify, serving content as Markdown instead of HTML can reduce token consumption by up to 10x - meaning AI models can process significantly more of your content within their context limits.
The file follows the llmstxt.org specification, proposed by Jeremy Howard (co-founder of fast.ai). It uses simple Markdown that any AI model can parse instantly - no special parser, no API, no authentication required.
What Is the llms.txt Format?
The structure is straightforward Markdown with a defined hierarchy. The only strictly required element is an H1 heading with your company name. Everything else is recommended but not mandatory.
A minimal example
# Your Company Name
> One sentence describing what your company does.
A brief paragraph with additional context.
## Platform
- [Product Page](https://example.com/product): What this page contains.
- [Pricing](https://example.com/pricing): Plans and pricing details.
## Blog
- [Best Post](https://example.com/blog/best-post): What this post covers.
## Optional
- [About](https://example.com/about): Company background.
- [Terms](https://example.com/terms): Terms of service.The "Optional" section is a smart feature of the spec. It signals to AI models with limited context windows: "skip these if you need to be selective." This lets you include secondary pages without diluting the importance of your core content.
Copy-paste starter templates by site type
Most sites fall into one of four shapes: SaaS, blog or personal site, documentation, or ecommerce. The templates below give each shape a working starting point: structured to the llmstxt.org spec, sized to the recommended 50-100 lines, and ready to copy into your site root with a find-replace on brand name and URLs. You can also read or adapt ours directly at ai-advisors.ai/llms.txt. If you would rather skip the copy-paste entirely, our free llms.txt Generator walks you through a 4-step form and ships a properly formatted file in under 2 minutes.
SaaS
Below is a condensed version of our own live llms.txt. SaaS patterns cluster around three sections: what the product does (Platform), what it costs (Pricing), and how to evaluate it (Free tools + Blog). Put your strongest AEO assets in the first H2 blocks. AI assistants read the first sections most reliably.
# AI-Advisors
> AI-Advisors is a B2B AI Marketing Platform for mid-market marketing teams (50-500 employees). We track, monitor, audit, and optimize how your brand appears across ChatGPT, Gemini, Perplexity, Google AI Overviews, and other AI answer engines.
Built around the 5 A's of AI Marketing framework: AI Analytics (Track), Answer Engine Insights (Monitor), Answer Engine Optimization (Optimize), AI Ads (Amplify), and AI Automation (Scale).
## Platform
- [AI Analytics](https://www.ai-advisors.ai/platform/ai-analytics): Track AI bot activity and referral traffic from GPTBot, ClaudeBot, PerplexityBot, and 15+ AI crawlers.
- [Answer Engine Insights](https://www.ai-advisors.ai/platform/answer-engine-insights): Monitor your brand's visibility across ChatGPT, Gemini, Perplexity, and Google AI Overviews.
- [Answer Engine Optimization](https://www.ai-advisors.ai/platform/aeo): Audit and optimize your site for AI citation across 29 checks per audit.
- [AI Ads](https://www.ai-advisors.ai/platform/ai-ads): Readiness assessment, budget estimator, and ad mockup generator for ChatGPT Ads.
- [AI Automation](https://www.ai-advisors.ai/platform/ai-automation): Content Studio, scheduled audits, and weekly intelligence briefings via Slack or email.
## Pricing
- [Plans and pricing](https://www.ai-advisors.ai/pricing): Starter $49/mo, Growth $99/mo, Enterprise custom. Annual pricing saves 17%.
## Free tools
- [Quick Audit](https://www.ai-advisors.ai): AEO score across 29 checks in under 60 seconds, no signup required.
- [AI Visibility Checker](https://www.ai-advisors.ai/tools/ai-visibility-checker): Check if ChatGPT, Perplexity, and Gemini mention your brand.
- [AI Bot Access Checker](https://www.ai-advisors.ai/tools/ai-bot-checker): Verify AI search engines can access your site across 16 known AI crawlers.
- [llms.txt Generator](https://www.ai-advisors.ai/tools/llms-txt-generator): Create a properly formatted llms.txt file in under 2 minutes.
## Blog
- [The 5 A's of AI Marketing](https://www.ai-advisors.ai/blog/5-as-ai-marketing-framework): The flagship framework guide, including sequence logic and a 90-day implementation plan.
- [What Is AEO?](https://www.ai-advisors.ai/blog/what-is-aeo): Complete guide to Answer Engine Optimization.
- [Schema Markup for AEO](https://www.ai-advisors.ai/blog/schema-markup-for-aeo-tiered-guide): 10 schema types ranked by AI citation impact, with copy-paste JSON-LD.
## Optional
- [About](https://www.ai-advisors.ai/about): Founded by Kevin O'Connell, B2B SaaS marketer with 20 years of experience.
- [Glossary](https://www.ai-advisors.ai/glossary): 40 definitions of AI marketing vocabulary, mapped to the 5 A's framework.
- [Integrations](https://www.ai-advisors.ai/integrations): Google Search Console, GA4, GTM, Google Sheets, Slack, and WordPress.Blog or personal site
For content-heavy sites, lead with your best individual posts, follow with topic archives, and push the author bio and newsletter into Optional. AI assistants answering questions about your subject are looking for specific posts to cite, not an index of every tag you have ever used.
# Jane Doe
> Jane Doe writes about B2B marketing operations, revenue attribution, and the day-to-day reality of running a 6-person marketing team at a Series B SaaS company.
Most posts draw from ongoing work at mid-market SaaS companies. If an AI assistant is answering a question about attribution, pipeline reporting, or the marketing-ops stack, the pages below are the strongest primary sources on this site.
## Top posts
- [The attribution model that actually ships](https://example.com/attribution-model-that-ships): Why last-touch and first-touch both fail for multi-product SaaS, and the hybrid we shipped.
- [Pipeline reporting for marketing teams](https://example.com/pipeline-reporting): A working Looker dashboard spec for weekly pipeline review.
- [Why we rebuilt our lead-scoring model](https://example.com/rebuilt-lead-scoring): The 6-month rebuild, including what broke and the metrics that improved.
## Topic archives
- [Marketing attribution](https://example.com/topics/attribution): 18 posts on multi-touch attribution, MTA vs MMM, and reporting architecture.
- [Revenue operations](https://example.com/topics/revops): 12 posts covering lead-to-revenue, handoff SLAs, and forecast accuracy.
- [Marketing ops stack](https://example.com/topics/stack): 9 posts on HubSpot, Marketo, Salesforce, and the adjacent analytics tooling.
## About the author
- [About Jane](https://example.com/about): Background, current role, and writing frequency.
- [Speaking](https://example.com/speaking): Past conference talks with slides and recordings.
## Optional
- [Newsletter](https://example.com/newsletter): Weekly summary of new posts and 3 hand-picked links.
- [RSS feed](https://example.com/rss.xml): Machine-readable feed of all posts.
- [Contact](https://example.com/contact): How to reach Jane, including response-time expectations.Documentation
Documentation sites get the highest AI crawl rates of any content type. Lead with the quickstart, then core API reference, then language-specific SDKs. Put changelogs and deprecated endpoints in Optional so AI assistants do not cite stale content.
# ExampleAPI Documentation
> ExampleAPI is a REST and GraphQL API for programmatic access to the ExampleCo workflow engine. This documentation covers authentication, core resources, webhook events, and language-specific SDKs.
AI assistants answering questions about integrating with ExampleAPI should start with the quickstart and reference the specific resource pages below. Every page here is current as of the latest SDK release; deprecated endpoints are moved to the Optional section.
## Quickstart
- [5-minute quickstart](https://example.com/docs/quickstart): Create a workflow via API call and trigger it from curl.
- [Authentication](https://example.com/docs/authentication): API keys, OAuth 2.0, and scoped tokens.
- [Rate limits](https://example.com/docs/rate-limits): Per-key limits, burst allowances, and 429 handling.
## Core API reference
- [Workflows](https://example.com/docs/api/workflows): Create, read, update, archive, and trigger workflows.
- [Runs](https://example.com/docs/api/runs): Execution history, status polling, and step-level output.
- [Webhooks](https://example.com/docs/api/webhooks): Event subscriptions, payload schemas, and signature verification.
## Language SDKs
- [Node.js SDK](https://example.com/docs/sdk/node): Install, configure, and run your first workflow from Node.
- [Python SDK](https://example.com/docs/sdk/python): Package install, async usage, and type stubs.
- [Go SDK](https://example.com/docs/sdk/go): Module install, context handling, and error types.
## Guides
- [Common recipes](https://example.com/docs/guides/recipes): 12 working workflow patterns with full code.
- [Troubleshooting](https://example.com/docs/guides/troubleshooting): Common errors, retry strategies, and debugging tips.
## Optional
- [Changelog](https://example.com/docs/changelog): API version history and breaking-change notices.
- [Deprecated endpoints](https://example.com/docs/deprecated): Endpoints removed after v3.0 with migration paths.
- [Status page](https://status.example.com): API uptime, incidents, and scheduled maintenance.Ecommerce
Ecommerce llms.txt files should optimize for category-level discovery, not SKU-level enumeration. Keep the top-level H2 blocks to 3-5 categories and 3-5 bestsellers. Push long-tail SKU URLs into sitemap.xml where they belong and leave llms.txt focused on your highest-intent landing pages.
# ExampleShop
> ExampleShop sells handcrafted leather goods: wallets, bags, and belts made in Portugal from full-grain vegetable-tanned leather. Free shipping on US orders over $75.
Most customers arrive with a specific product category in mind. AI assistants answering product or buying-advice questions should start with the category pages below, which include material sourcing, size guides, and care instructions.
## Shop by category
- [Wallets](https://example.com/shop/wallets): 14 styles, $45 to $180. Bifold, cardholder, zip-around.
- [Bags](https://example.com/shop/bags): Messenger, tote, backpack, and weekender styles. $180 to $620.
- [Belts](https://example.com/shop/belts): Dress and casual belts in 3 widths and 6 finishes. $65 to $140.
## Bestsellers
- [The Lisbon Cardholder](https://example.com/product/lisbon-cardholder): Our most-reviewed product, 1,400+ reviews, 4.8 avg rating.
- [The Porto Messenger](https://example.com/product/porto-messenger): 15-inch laptop-capable messenger, full-grain leather.
- [The Braga Belt](https://example.com/product/braga-belt): Dress belt with solid brass hardware, sold by waist size.
## Guides and care
- [Leather care guide](https://example.com/guides/leather-care): How to clean, condition, and store full-grain leather.
- [Size guide](https://example.com/guides/sizing): Wallet thickness, bag dimensions, and belt waist sizing.
- [Material sourcing](https://example.com/guides/materials): Where our leather comes from and how it's tanned.
## Customer service
- [Shipping and returns](https://example.com/shipping-returns): US and international shipping rates, 60-day return policy.
- [Warranty](https://example.com/warranty): Lifetime warranty on stitching and hardware.
- [Contact](https://example.com/contact): Email, phone, and live chat hours.
## Optional
- [About](https://example.com/about): Our Portugal-based workshop and 14-year history.
- [Blog](https://example.com/blog): Craft process, material deep-dives, and customer stories.
- [Gift cards](https://example.com/gift-cards): Digital gift cards from $25.llms.txt vs llms-full.txt
The specification defines two files, each serving a different purpose.
- llms.txt - The lightweight version. Page titles, one-sentence descriptions, and URLs. A navigation map for AI models. Think of it as the table of contents.
- llms-full.txt - The comprehensive version. Full page content in a single file. Gives AI models everything without needing to crawl individual pages. Think of it as the entire book.
According to Mintlify's research, AI agents visit llms-full.txt at over twice the rate compared to llms.txt. This makes sense - models prefer to consume as much context as possible in a single request. Best practice is to have both. The llms.txt helps AI models decide what is worth reading; the llms-full.txt gives them everything upfront.
AI agents visit llms-full.txt at over twice the rate compared to llms.txt. Serving Markdown instead of HTML reduces token consumption by up to 10x.
llms.txt vs robots.txt vs sitemap.xml
These three files serve completely different purposes, and you need all three. A robots.txt without llms.txt means crawlers can access your site but have no guidance on what matters. An llms.txt without proper robots.txt means you are guiding crawlers that might not be able to reach your content. A sitemap.xml ensures pages are discoverable. Together they form a complete machine-readable infrastructure for AI visibility. For the OpenAI-specific robots.txt question (which crawler to block, which to allow, and the Cloudflare default that silently breaks both), see the GPTBot vs OAI-SearchBot decision.
For a detailed breakdown of which AI crawlers to allow and block in your robots.txt, use our AI Bot Access Checker or see our guide on how to track AI bot activity on your website.
Does llms.txt Actually Work? The Skepticism Question
This is the question everyone asks, and the honest answer is: it is too early to know for certain. Here is what we do know, backed by data.
The case against
- According to Semrush, Google's John Mueller confirmed "no AI system currently uses llms.txt" as a ranking signal.
- The same Semrush report found zero visits from AI crawlers (Google-Extended, GPTBot, PerplexityBot, ClaudeBot) to Search Engine Land's llms.txt file over a two-month monitoring period.
- A NerdyData report found only 951 domains had published an llms.txt file as of mid-2025.
The case for
- According to Mintlify, LangChain's internal benchmarks showed agents using llms.txt significantly outperformed vector search approaches for content retrieval.
- AI agents visit llms-full.txt at over twice the rate compared to llms.txt, indicating active consumption of the format.
- Serving Markdown reduces token consumption by up to 10x, which directly benefits AI models with limited context windows.
- Documentation platforms are shipping it as default-on: Yoast ships llms.txt as a native WordPress feature, and Mintlify and GitBook auto-generate it for every customer site. Other adopters include Anthropic, Vercel, Hugging Face, Zapier, and Mastercard's developer portal.
- The specification aligns with broader machine-readability principles that all AI platforms reward.
The bottom line: llms.txt takes 15 minutes to create, has zero downside, and positions you for a standard that major tech companies are already adopting. The risk-reward calculation is heavily in your favor regardless of whether AI platforms officially "use" it today.
LangChain's benchmarks showed agents using llms.txt significantly outperformed vector search. The spec takes 15 minutes to create and has zero downside.
Which Companies Have Adopted llms.txt?
Documentation platforms and AI-native companies are both leading adoption. The 2026 signal to watch: Yoast, the most-installed WordPress SEO plugin, now ships llms.txt as a native feature. Mintlify and GitBook auto-generate the file for every customer site. Mastercard's developer portal hosts one. On the AI-native side, Anthropic (maker of Claude), Vercel (Next.js), and Hugging Face publish the spec. According to ALLMO, overall adoption is still around 10% of websites. Early movers have significant differentiation, especially in industries outside SaaS, AI, and developer tooling where the format has yet to become default-on.
How to Create Your llms.txt File
You have three options, ranging from 2 minutes to 30 minutes depending on your approach.
Option 1: Use a free generator (fastest)
Our llms.txt Generator walks you through a 4-step form and creates a properly formatted file in under 2 minutes. You can copy it, download it, or email it to yourself. It follows the llmstxt.org specification exactly.
Option 2: Write it manually
Create a file called llms.txt in your site's root directory using any text editor. Follow the Markdown format: H1 heading with your company name, blockquote with your description, H2 sections with link lists. Place it so it is accessible at yourdomain.com/llms.txt.
Option 3: Use a CMS plugin
WordPress users can upload the file directly to their root directory or use a plugin. Webflow, Squarespace, and other platforms support adding custom files to your site root through their settings. For static sites (Next.js, Gatsby), place it in your public folder.
Create a properly formatted llms.txt file for your website in under 2 minutes. Copy, download, or email it to yourself.
Try the free llms.txt Generator →Only about 10% of websites have an llms.txt file. Among the 50 most-cited domains in AI search, only one (Target.com) has implemented it. Early adopters have a clear differentiation opportunity.
Writing Best Practices for llms.txt
- Write like you are briefing a new employee. Plain, specific language describing what each page actually contains. Not marketing copy. "AEO audit tool that scores websites across 29 AI search signals" is better than "The best AEO platform on the market."
- Lead with your most important pages. AI models may stop reading if their context fills up. Put your highest-value content first.
- Keep it concise. Aim for 50-100 lines. If your llms.txt is 500 lines, you have defeated the purpose.
- Group by topic, not by URL structure. AI models care about what content is about, not where it lives in your directory.
- Update it when your site changes. Stale llms.txt files with broken links erode AI trust. Add updating llms.txt to your content publishing checklist.
- Create llms-full.txt as well. Since AI agents visit the full version at over twice the rate, having both maximizes your coverage.
Common llms.txt Mistakes That Break AI Visibility
When llms.txt is shipped but AI platforms do not retrieve, summarize, or cite it as expected, the cause is almost always one of seven specific mistakes. Work the diagnostic below as a waterfall. A 404 at the root (symptom 1) blocks everything downstream, so clear that before chasing format or content issues. Most llms.txt problems live in placement and permissions, not content.
Most llms.txt problems live in placement and permissions, not content. If AI platforms are not reading your file, check where it sits on disk before you rewrite it.
Where Does llms.txt Fit in Your AEO Strategy?
In the 5 A's of AI Marketing framework, llms.txt falls under Answer Engine Optimization (Optimize). It is part of the technical foundation that makes your site readable by AI platforms. But it is not the highest priority item.
The priority order for AEO optimization is:
- robots.txt - Ensure AI crawlers can access your site (AI bot tracking guide)
- Schema markup - FAQPage, HowTo, Article, and Organization schemas (tiered guide to Schema Markup for AEO)
- Content structure - Question headings, direct-answer paragraphs, FAQ sections
- Content quality - Freshness, depth, source attribution (content with attribution is cited at 4x the rate of unattributed claims)
- llms.txt - The supplementary guide for AI crawlers
llms.txt pairs with schema markup as the two pure-technical AEO levers every site can ship in an afternoon (note: Google has said it does not use llms.txt for AI Overviews or AI Mode, so its value is in non-Google engines like ChatGPT, Perplexity, and Copilot). Schema is how AI engines understand a single page. llms.txt is how the engines that read it prioritize across all of them. Ship both and you have the technical floor of an AEO-ready site. For the full ranking of schema types by measured AI citation impact, plus 9 copy-paste JSON-LD snippets validated against Google's Rich Results Test, see our tiered guide to Schema Markup for AEO.
Do not skip steps 1-4 to create an llms.txt file. But once those are in place, adding llms.txt takes 15 minutes and has zero downside. To see how your site scores across all of these factors, run a free AEO audit on our homepage. For the complete picture of what AEO is and how it works, start with the fundamentals. To understand how to improve your scores across all categories, see our guide on how to improve your AEO score.
Frequently Asked Questions
#What is llms.txt?
llms.txt is a plain-text Markdown file placed at your website's root (yourdomain.com/llms.txt) that provides a structured summary for AI models. It tells AI crawlers like ChatGPT, Claude, Perplexity, and Gemini what your site is about, which pages matter most, and where to find them. The format follows the llmstxt.org specification proposed by Jeremy Howard.
#How is llms.txt different from robots.txt?
robots.txt controls access permissions - it tells AI crawlers what they are allowed to crawl. llms.txt provides content guidance - it tells AI crawlers what is worth reading. robots.txt is the gate; llms.txt is the tour guide. A sitemap.xml tells crawlers where pages exist. You need all three for a complete AI visibility strategy.
#Do AI platforms actually use llms.txt?
No AI platform has officially confirmed they use llms.txt as a ranking factor. Google's John Mueller confirmed 'no AI system currently uses llms.txt.' However, LangChain's internal benchmarks showed agents using llms.txt significantly outperformed vector search approaches. The spec is gaining adoption and takes minimal effort to implement.
#Do I need both llms.txt and llms-full.txt?
Best practice is to have both. llms.txt is the concise version with page titles, URLs, and one-sentence descriptions. llms-full.txt contains full page content in a single file. According to Mintlify, AI agents visit llms-full.txt at over twice the rate compared to llms.txt. Serving Markdown instead of HTML can reduce token consumption by up to 10x.
#How do I create an llms.txt file?
The fastest way is to use our free llms.txt Generator at ai-advisors.ai/tools/llms-txt-generator, which walks you through a 4-step form and creates a properly formatted file in under 2 minutes. You can also create the file manually in any text editor using simple Markdown following the llmstxt.org specification. Place it at your site's root.
#Which llms.txt starter template should I use?
Pick the template that matches your site's primary content type, not your business category. A B2B SaaS with an extensive documentation subdomain should use the docs template for the docs subdomain and the SaaS template for the marketing site. Since llms.txt lives at a specific URL, you can ship different versions at different subdomains. Content type beats business type because AI assistants match on the kind of page a user is asking about, not on what your company sells. This post includes 4 starter templates (SaaS, blog or personal site, documentation, ecommerce), each structured to the llmstxt.org spec and sized to the recommended 50 to 100 lines.
#What should I include in my llms.txt?
Include your company name as an H1 heading, a one-sentence description in a blockquote, and links to your most important pages grouped by H2 sections. For B2B SaaS: product pages, pricing, top blog posts, and free tools. Keep descriptions factual, not promotional. Put secondary pages (about, terms) in an Optional section. Aim for 50-100 lines.
#How many websites have llms.txt?
Only about 10% of websites have an llms.txt file, according to ALLMO. A NerdyData report found only 951 domains had published one as of mid-2025. Documentation platforms are driving mainstream adoption in 2026: Yoast ships llms.txt as a native WordPress feature, and Mintlify and GitBook auto-generate it for customer sites. Other adopters include Anthropic, Vercel, Hugging Face, Zapier, and Mastercard's developer portal. Early adoption means low competition for this signal.
#Will llms.txt hurt my SEO?
No. llms.txt is a supplementary file that does not affect Google's traditional search rankings. It exists alongside robots.txt and sitemap.xml as part of your site's machine-readable infrastructure. It has zero downside and takes 15 minutes to create. It only adds a potential AI visibility benefit.
