What Is llms.txt and Should Your Site Have One?

llms.txt is a plain-text file at the root of your website that tells AI models what your site is and which pages matter. It was proposed as an open standard at llmstxt.org. This covers what goes in one, whether it is worth adding, and how to avoid the common mistakes.

What is llms.txt?

llms.txt is a Markdown file served at yourdomain.com/llms.txt that gives language models a curated summary of your site. It typically contains a one-line description of what you do, a short paragraph of context, and a linked list of your key pages grouped by section.

The reasoning behind it: an AI model working from a raw crawl has to infer what matters from navigation, internal links and page volume. That inference is often wrong. A small site with one genuinely important page can look trivial; a large site can bury its best content under hundreds of thin pages. llms.txt lets you state it directly.

How is it different from robots.txt and sitemap.xml?

They answer three different questions:

File	Question it answers	Audience
robots.txt	What am I allowed to crawl?	All crawlers
sitemap.xml	What URLs exist?	Search engines
llms.txt	What is this site about and what matters?	Language models

sitemap.xml is exhaustive and machine-generated. It lists everything, weighted equally. llms.txt is the opposite: short, curated, opinionated. You are choosing what to put in front of the model rather than dumping the whole index.

You want all three. They do not overlap.

What goes in an llms.txt file?

The proposed format is deliberately simple Markdown:

1. An H1 with your site or business name
2. A blockquote with a one-line summary
3. Optional context paragraphs explaining what the site covers
4. H2 sections grouping links, each as [Title](url): short description

A trimmed example:

# Resulve

> AI-powered marketing agency in the Philippines.

Content is structured for both human readers and AI tools: blog posts
include TL;DRs and FAQs, service pages describe deliverables and pricing.

## Services

- [Paid Media](https://resulve.app/services/paid-media): Google, Meta and
  ChatGPT ads that put budget where the buyers are.

## Writing

- [AI Visibility in the Philippines](https://resulve.app/blog/ai-visibility-philippines):
  How to get your business cited by ChatGPT, Perplexity and Google AI.

You can see the full live version at resulve.app/llms.txt.

Should you actually bother?

Here is the honest position. No AI engine currently guarantees it reads your llms.txt. It is a community proposal, not a specification published by OpenAI, Google or Anthropic. Anyone telling you it is required is overselling it.

The case for adding one anyway is the asymmetry. It takes about an hour. If adoption grows, you are already there. If it does not, you lost an hour and gained a genuinely useful exercise: being forced to write one sentence describing what your business does, and choosing which ten pages actually matter, is clarifying in its own right.

What it will not do is improve your Google rankings. It has no effect there. Treat it as AI visibility work, which is a different objective. The distinction is covered in GEO vs SEO.

How do you generate it without maintaining it by hand?

Hand-writing it once is fine. Keeping it current by hand is where it rots.

The better pattern is generating it from the content you already have, the same way you generate sitemap.xml. If your services and posts live in structured files or a database, the file can be assembled at request time so new content appears automatically and deleted content disappears.

That is how the example above works. It is generated from the same source that drives the sitemap and the navigation, so it can never drift out of date.

Three mistakes worth avoiding:

• Listing every page. That is what sitemap.xml is for. Curate to what matters.
• Writing marketing copy in it. Models are extracting facts, not being persuaded. Say what things are.
• Letting it go stale. A file listing pages that 404 is worse than no file at all.

Frequently Asked Questions

What to do next

If you already have sitemap.xml and a sensible robots.txt, adding llms.txt is a short job that completes the set. Then move on to the work that carries more weight: direct answers, clean structure and schema markup for AI search.

The full sequence is in our guide to AI visibility in the Philippines.