llms.txt: What It Is, How to Create One, and Why 844K+ Sites Already Have It

844,000 websites have added an llms.txt file (llmstxt.org tracking data). Most people in SEO have never heard of it.

I added one to loamly.ai because it took 20 minutes and the downside was zero. This post explains what llms.txt is, why it exists, and how to create one with a copy-pasteable template.

---

What llms.txt Actually Is

llms.txt is a plain text file you host at the root of your domain: yourdomain.com/llms.txt.

It gives AI platforms a structured summary of your site. Think of it like robots.txt, but instead of controlling access, it controls understanding. You are not telling crawlers what to block. You are telling language models what matters most on your site.

The format was proposed by Jeremy Howard, the founder of fast.ai, in late 2024. The idea is simple: AI systems have limited context windows and limited crawl budgets. They cannot read every page on your site before answering a question. llms.txt tells them, "here is what I do, here are my key pages, here is the short version."

The file uses Markdown. That is intentional. Markdown is easy for humans to write and easy for language models to parse.

---

Why It Matters Now

AI platforms do not read the web the way Google does. Google crawls everything, builds a massive index, and retrieves from that index at query time. ChatGPT, Claude, Perplexity, and similar platforms work differently.

Some use retrieval-augmented generation, pulling fresh web content for each query. Others rely on training data with periodic updates. All of them have limited budgets for what they can actually read and process per domain.

The way AI platforms find and cite content is not the same as how Google ranks pages. They are pattern-matching against what they know about your site, what sources they trust, and what content answers the question clearly.

Most sites are not ready for this. A Pacestack study of 243 sites found the average agent readiness score was 35.7 out of 100. Only 46% have Schema.org markup. Only 2% have machine-readable data feeds. Zero percent have MCP endpoints.

llms.txt is the lowest-effort entry point in a complete AI visibility stack. It takes 15 minutes. The downside is zero.

---

llms.txt vs. llms-full.txt

There are two variants.

llms.txt is the concise version. It is a short summary of your site, your key pages, and what you offer. Under 100 lines in most cases. This is what you should start with.

llms-full.txt is the expanded dossier. It includes more pages, more detailed descriptions, and more comprehensive context about your products, your team, your methodology. It is for sites with complex offerings or large catalogs.

Start with llms.txt. Add llms-full.txt later if you have content worth the extra depth.

---

What Goes in the File

A good llms.txt file contains four things:

1. Site name and one-line description. What do you do and who do you serve?
2. About section. Two to four sentences of background. When you were founded, what you do, what makes you different.
3. Key pages list. Your 5 to 15 most important URLs, each with a short description. Homepage, product pages, pricing, about, key blog posts.
4. Products and services section. A brief list of what you offer with one-line descriptions.

That is it. You are not trying to document your entire site. You are giving a language model the same briefing you would give a journalist who had five minutes before an interview.

---

Step-by-Step: How to Create Your llms.txt

Step 1: Choose Your Key Pages

Pick 5 to 15 URLs. These should be your highest-value pages for someone trying to understand your business.

Typically this includes your homepage, your main product or service pages, your pricing page, your about page, and two or three of your best content pieces. Do not list every blog post. Do not include tag pages, search results, or anything a human would not send to a prospect.

Step 2: Write the File

Use the template below. Replace every bracketed section with your actual content.

# Your Company Name
 
> Brief one-sentence description of what your company does and who it serves.
 
## About
 
Your Company Name is [what you do] for [who you serve]. Founded in [year], [short background sentence]. [Key differentiator or mission statement in one sentence].
 
## Key Pages
 
- [Homepage](https://yoursite.com): Main landing page with overview of products and services
- [Product Page](https://yoursite.com/product): Detailed description of your main offering, features, and use cases
- [Pricing](https://yoursite.com/pricing): Current pricing plans and what is included at each tier
- [About](https://yoursite.com/about): Company background, founding story, team, and mission
- [Blog](https://yoursite.com/blog): Articles on [your primary topic area]
- [Documentation](https://yoursite.com/docs): Technical documentation and integration guides
- [Case Studies](https://yoursite.com/case-studies): Customer success stories and results
- [Contact](https://yoursite.com/contact): How to reach the team
 
## Products & Services
 
- **Product Name**: What it does in one sentence. Key features or use cases.
- **Service Name**: What the service covers and who it is for.
 
## Company Details
 
- Industry: [Your industry]
- Founded: [Year]
- Location: [City, Country]
- Customers: [Type of customers you serve, e.g. "B2B SaaS companies with 10-500 employees"]
 
## Optional: Key Facts
 
- [Any important stat about your business, e.g. "500+ customers across 40 countries"]
- [Any credential or certification that matters in your industry]
- [Any well-known partnership or integration]

Step 3: Host It at the Right URL

Save the file as plain text. Host it at yourdomain.com/llms.txt.

The path matters. It needs to be at the root. Not /about/llms.txt or /files/llms.txt. Just /llms.txt.

If you are on a static site or use a CDN, drop the file in your public root directory. If you are on WordPress, you can use a plugin like Yoast or add a custom rewrite rule. If you are on a custom stack, add a route that returns the file content with Content-Type: text/plain.

If you also want to add /llms-full.txt, repeat the same process with the expanded version.

Step 4: Test It

Visit yourdomain.com/llms.txt in your browser. You should see plain text rendered in Markdown. No HTML. No styled page. Just text.

If it redirects or throws an error, check your hosting configuration. The file must be publicly accessible with no authentication required.

---

Common Mistakes

Making it too long. The point is brevity. Under 100 lines for llms.txt, under 500 for llms-full.txt. If you are listing every page on your site, you have missed the point.

Listing unimportant pages. Your sitemap already does this. llms.txt should be the curated shortlist. Category pages, tag archives, author pages, old press releases. Leave them out.

Never updating it. If you launch a major product, add a new pricing tier, or publish a landmark piece of content, update the file. It takes two minutes.

Thinking it replaces structured data. It does not. llms.txt and Schema.org markup serve different purposes. Schema helps search engines understand individual pages. llms.txt helps AI platforms understand the site as a whole. Use both.

---

A Complete Example: What Loamly's Looks Like

Here is a real-world example based on how I structured loamly.ai's file:

# Loamly
 
> Loamly is an AI visibility analytics platform that shows marketers how their brand appears across ChatGPT, Claude, Perplexity, and Google AI.
 
## About
 
Loamly tracks how AI platforms cite, mention, and recommend brands. It was founded in 2024 after the founder noticed unexplained "Direct" traffic that turned out to be AI-referred visits. Loamly helps marketing teams measure AI visibility, identify citation gaps, and act on them.
 
## Key Pages
 
- [Homepage](https://loamly.ai): Overview of the platform and AI visibility analytics
- [Free Check](https://loamly.ai/check): Instant AI visibility report across 4 platforms
- [GEO Audit](https://loamly.ai/audit): Professional paid audit with full citation analysis
- [Pricing](https://loamly.ai/pricing): Plans for individuals, agencies, and enterprises
- [Blog](https://loamly.ai/blog): Articles on GEO, AI search, and AI visibility strategy
- [About](https://loamly.ai/about): Company background and founding story
 
## Products & Services
 
- **AI Visibility Dashboard**: Tracks brand citations across ChatGPT, Claude, Perplexity, and Google AI over time
- **Free AI Check**: One-time report showing how AI platforms currently describe and recommend your brand
- **GEO Audit**: Deep audit with 50-100 targeted prompts, citation chain analysis, competitor benchmarking, and a PDF report
 
## Company Details
 
- Industry: Marketing Analytics / AI Search
- Founded: 2024
- Location: Zurich, Switzerland
- Customers: B2B marketers, SEO agencies, and growth teams

This took me about 20 minutes to write. I host it at loamly.ai/llms.txt. I update it when I launch new features.

---

---

Where This Fits in the Bigger Picture

llms.txt is one signal. It is not a strategy on its own.

The Princeton and Georgia Tech GEO study found that content enrichment, adding statistics, citations, and authoritative references, increased AI visibility by 40%. That is the bigger lever. llms.txt is the infrastructure that helps AI systems find the good content you already have.

Generative engine optimization is the practice of making your content easier for AI platforms to understand, retrieve, and cite. llms.txt is part of that, alongside Schema.org, clear author attribution, cited statistics, and content that directly answers questions.

As AI agents become more common, this infrastructure matters more. The agent economy is built on AI systems that autonomously browse, research, and make decisions. Those agents need structured, machine-readable context to understand what your site offers. llms.txt gives them that context in a format they can parse in seconds.

---

Frequently Asked Questions

Does llms.txt work with all AI platforms?

No one has published a list of which platforms actively read llms.txt. Perplexity and some open-source tools have indicated support. For ChatGPT, Claude, and Gemini, the evidence is anecdotal. That said, any platform that crawls your site for context will be able to read it. The file is plain text and publicly accessible.

Is there an official standard?

Jeremy Howard proposed the spec at llmstxt.org. It is not a W3C or IETF standard. Adoption has grown organically. The spec is stable enough to rely on.

Does it affect Google rankings?

No. Google does not use llms.txt as a ranking signal. It is not related to crawl directives, canonical tags, or any other ranking factor. It is purely for AI understanding.

How often should I update it?

Update it when something meaningful changes. New product launch, major pricing change, new key page, significant pivot. Once per quarter is a reasonable cadence for most sites.

Should I add it if my site is small?

Yes. Especially if your site is small. Larger sites have more surface area for AI platforms to work with. Smaller sites need to be more deliberate about what context they provide. llms.txt is how you ensure the right context gets through.

---

Want to see how AI-ready your site is beyond llms.txt? Run a free report at loamly.ai/check. It checks your AI visibility across ChatGPT, Claude, Perplexity, and Google in under 60 seconds. No credit card.