Set up llms.txt in 5 minutes and get cited by ChatGPT

You have a robots.txt. You probably have a sitemap.xml. Now there's a third file AI crawlers are looking for — and most websites don't have it yet.

llms.txt is an emerging standard that gives AI assistants a plain-language summary of your site: what it is, who it's for, and which pages matter most. Think of it as a README for language models — a concise brief that helps ChatGPT, Perplexity, and other AI systems understand your content structure without having to crawl every page.

The file takes about five minutes to create. Here's exactly how to do it.

What llms.txt is (and isn't)

llms.txt was proposed by Jeremy Howard in 2024 as a lightweight convention — not an official standard yet, but increasingly supported by AI platforms. It lives at the root of your domain (https://yourdomain.com/llms.txt) and contains a brief, structured description of your site in Markdown format.

It is not a replacement for robots.txt. robots.txt controls crawler access (what bots are allowed to crawl). llms.txt controls context (what AI systems should understand about your site). Both serve different purposes and you need both.

The llms.txt format

The format is simple Markdown with a few conventional sections. Here's the structure:

# Your Site Name

> One-sentence description of what your site is and who it's for.

## About
Brief paragraph (2–4 sentences) explaining your site's purpose,
primary audience, and what makes it authoritative on the topics
you cover.

## Key Pages
- [Homepage](https://yourdomain.com/): What the main page covers
- [Feature/Product](https://yourdomain.com/product/): Key offering
- [Blog](https://yourdomain.com/blog/): Content categories covered

## Topics
Comma-separated list of topics this site is authoritative about.

## Contact
Optional: contact information or about page link.

A real example: CiteReady's llms.txt

Here's the actual llms.txt for this site:

# CiteReady

> CiteReady is a free GEO audit tool that checks how well a website
> is optimized for AI search (ChatGPT, Perplexity, Google AI Overviews).

## About
CiteReady analyzes any URL and returns a GEO score across four
categories: AI Crawler Access, Content Citability, Structured Data,
and Technical Foundation. It is built by SpryTools and available
at citeready.sprytools.com.

## Key Pages
- [GEO Audit Tool](https://citeready.sprytools.com/): Free audit,
  no signup required
- [Blog](https://citeready.sprytools.com/blog/): Guides on GEO,
  llms.txt, robots.txt for AI crawlers, and structured data

## Topics
Generative Engine Optimization, GEO, AI search visibility, llms.txt,
robots.txt AI crawlers, structured data, ChatGPT citations,
Perplexity citations, Google AI Overviews

Step-by-step: setting up llms.txt

1. Draft your summary Write one sentence that describes what your site does and who it's for. This is the most important line — it's often what AI systems extract when summarizing your site.
2. List your key pages Pick 5–10 pages that best represent your site. Include your homepage, main product/service pages, and your top content pages. Write a short description next to each link.
3. Define your topic authority List the topics you want to be cited for. Be specific. "Marketing" is too broad; "B2B email marketing automation" tells the AI exactly when to cite you.
4. Save as plain text at /llms.txt Create the file as plain UTF-8 text and place it at the root of your domain. Serve it as text/plain — no HTML wrapper needed.
5. Verify with a GEO audit Run a CiteReady audit on your homepage. The Technical Foundation section checks whether llms.txt exists, is reachable, and contains a valid description.

Common mistakes to avoid

• Serving it as HTML. Some servers convert all requests to HTML. Make sure /llms.txt returns Content-Type: text/plain, not a 301 redirect to an HTML page.
• Making it too long. Keep it under 500 words. The point is conciseness — a wall of text defeats the purpose. AI systems will read your actual pages for depth; llms.txt is for orientation.
• Listing every page. Prioritize. Listing 200 URLs is noise. List the 5–10 pages that best represent your site's core value.
• Ignoring the description line. The > description line under the heading is the most valuable part of the file. Don't skip it or make it generic ("Welcome to our website").
• Blocking it in robots.txt. If you have Disallow: /*.txt in your robots.txt, AI crawlers can't read your llms.txt. Check that /llms.txt is explicitly allowed.

Does llms.txt actually help with citations?

The short answer: yes, especially for newer or smaller sites where AI systems have less training data to work from. A clear llms.txt reduces the cognitive load on the AI — it doesn't have to infer what your site is about from scattered page content. It knows immediately, and that reduces the risk of misrepresentation or omission.

For established sites with high domain authority, the marginal benefit is smaller but still meaningful — particularly for Perplexity, which is known to crawl llms.txt files actively as part of its real-time indexing pipeline.