[becitable]
[Technical Foundation]

llms.txt: what it is, why it matters, and how to add it

Category: Technical Foundation · Reading time: 6 min

A small folded card on a stone surface, symbolising a simple discovery file for AI tools

llms.txt is one of the most talked-about and least understood files in GEO. The truth is somewhere between the breathless takes ("it's the new robots.txt") and the dismissive ones ("nobody uses it"). Here's what it actually is, and what to do about it.

What is llms.txt?

llms.txt is a plain text file placed at the root of your domain (e.g. becitable.com/llms.txt) that summarises your site for AI tools. It's written in Markdown, lists your most important pages, and gives a short description of what your site is about. Think of it as a curated table of contents aimed at language models.

How is llms.txt different from robots.txt?

robots.txt tells crawlers what they're allowed to access. llms.txt tells AI tools which pages you most want them to read. One is about permission, the other is about attention. They're complementary, not alternatives.

Do AI tools actually honour llms.txt?

As of mid-2026, the honest answer is "not really, not yet." The major commercial AI tools - OpenAI, Anthropic, Google, Perplexity - have not publicly committed to honouring llms.txt as a crawl directive. They still rely primarily on their existing web indexes.

That said: the file is harmless, costs nothing to publish, and is being adopted as a convention by AI-friendly tools and agents. If and when adoption hardens, you'll already be ahead. For small SaaS teams, the upside is asymmetric.

Should B2B SaaS sites publish an llms.txt?

Yes, with the right expectations. Treat it as low-cost insurance, not a foundational GEO move. Publish a good one, link to it from your sitemap, then forget about it and focus on the things that actually move citations today: schema, content, off-site authority.

What goes in an llms.txt file?

The conventional structure is:

  1. An H1 with your site name.
  2. A blockquote with a one-line description of what your site is.
  3. One or more H2 sections listing key pages as Markdown links, each with a brief description.
  4. Optional sections for topics covered, brand assets, or contact.
Example llms.txt
# Acme Analytics

> Product analytics for B2B SaaS teams. Helps PMs and growth teams track activation, retention and revenue without writing SQL.

## Core pages
- [Home](https://acmeanalytics.com): Overview of Acme Analytics
- [Product](https://acmeanalytics.com/product): What Acme does, features and integrations
- [Pricing](https://acmeanalytics.com/pricing): Plans, seats and usage limits
- [Docs](https://acmeanalytics.com/docs): Developer documentation and API reference

## Topics covered
- Product analytics for SaaS
- Activation, retention and revenue metrics
- Warehouse-native analytics (Snowflake, BigQuery)
- Integrations with HubSpot, Segment, Salesforce

How do I publish llms.txt on my site?

  1. Create a file named llms.txt in your site's public directory (e.g. /public/llms.txt).
  2. Write it in Markdown using the structure above. Keep it under 100 lines.
  3. Deploy and confirm it's reachable at yourdomain.com/llms.txt with a 200 response.
  4. Optionally link to it from your homepage footer for human discoverability.

Update it any time you launch a major new page or section. Otherwise it can sit untouched for months.

What llms.txt mistakes should I avoid?

  • Listing every URL on your site. Curate. Aim for the 10-20 pages you'd want AI to ground on.
  • Vague descriptions ("our pricing page"). Write the description the way you'd want it quoted back.
  • Forgetting to update it after a rebrand or major IA change.
  • Treating it as a substitute for schema. It isn't.

Want the bigger picture? See technical foundation for GEO or how to make your SaaS site crawlable for AI.