Diagnose your hotel

llms.txt for Hotels: What It Is, and Isn't

A plain-text file for AI agents, genuinely new, genuinely low-effort, and genuinely not a ranking guarantee. What it is, and isn't.

llms.txt is a plain-text file, published at yourdomain.com/llms.txt, that summarizes a website in a format AI agents can read directly, an emerging, not-yet-standardized convention, low-effort to publish, but with no evidence yet that it improves whether an AI engine names or ranks a hotel.

Key takeaways

  • ✓llms.txt is a real, simple text file, not a technical integration, most hotels can publish one in under an hour.
  • ✓It's genuinely new and non-standard: no major AI engine has confirmed it factors into ranking or selection decisions.
  • ✓It doesn't replace robots.txt, schema.org markup, or crawler access, it's a small addition alongside those, not a substitute.
  • ✓A copy-paste template and a real crawler checklist are both included below.
  • ✓Staylight publishes its own real, live llms.txt, linked below, as a working example.

What llms.txt actually is

llms.txt is a plain-text file published at a predictable location, yourdomain.com/llms.txt, that gives an AI agent a short, structured summary of a site: what the business is, its main pages, and a one-line description of each. It's similar in spirit to robots.txt (a simple, plain-text file crawlers can check), but where robots.txt controls access, llms.txt is closer to a hand-written index, meant to help an AI agent understand a site faster than crawling every page itself.

The proposal is genuinely new (first floated in 2024, still evolving), and it isn't an official web standard the way robots.txt is. No major AI company has published confirmation that it factors into how ChatGPT, Gemini, Claude, or Perplexity choose which hotels to name.

What it isn't

It isn't a ranking signal with proven effect, publishing one doesn't guarantee (or even demonstrably improve) whether an AI engine names your hotel. It isn't a replacement for robots.txt, which actually controls crawler access, or for schema.org markup, which gives AI engines structured, machine-readable facts. And it isn't urgent: a hotel missing llms.txt today isn't doing anything wrong, it's simply not yet ahead of an emerging convention.

Treat it as a small, low-effort addition once the more consequential items (crawler access, structured data, NAP consistency) are already handled, not as a priority ahead of them.

A copy-paste template for a hotel site

# [Hotel Name]

> [One or two sentences: what the hotel is, where it is, and what makes it distinct.]

## Rooms & Rates
- [Link to rooms/rates page]: [one-line description]

## Amenities
- [Link to amenities page]: [one-line description]

## Location
- [Link to location/directions page]: [one-line description]

## Booking
- [Link to booking engine]: [one-line description]

## Contact
- [Phone, email, or contact page link]

Fill in the brackets with real, current details, and keep each line genuinely short, this is meant to be scanned quickly, not read as prose. Publish it at yourdomain.com/llms.txt as a plain text file, no special server configuration needed beyond making sure the file is actually reachable at that exact path.

Which crawlers to actually allow

llms.txt doesn't control crawler access on its own, robots.txt does, so publishing an llms.txt file is only useful if the same AI crawlers can already reach your site. The real, current crawler tokens worth explicitly allowing in robots.txt: GPTBot, OAI-SearchBot, and ChatGPT-User (OpenAI), ClaudeBot, Claude-SearchBot, and Claude-User (Anthropic), Google-Extended (Google), and PerplexityBot and Perplexity-User (Perplexity). See the crawler readiness guide linked below for the full checklist and how to verify your own robots.txt isn't blocking any of them.

llms.txt vs. llms-full.txt

A related, less common convention is llms-full.txt, a fuller, full-text companion covering more of a site's actual content rather than just a short index. It's worth publishing only after the shorter llms.txt exists and only if it can be generated from the same real content that renders the site, a hand-maintained full-text file is more likely to go stale than a short index is.

A real, working example

Staylight publishes its own real, live llms.txt at staylight.io/llms.txt, generated directly from the same content that renders the site's Learn, Compare, and For-hotels pages, rather than a hand-maintained file that can drift out of date. It's a genuine, checkable example of the format in practice, not a hypothetical.

Questions

Will publishing llms.txt improve my hotel's AI visibility score?

There's no published evidence it does, from any major AI company. It's a low-effort, forward-looking step worth taking once the higher-impact items (crawler access, structured data, listing consistency) are handled, not a substitute for them.

Is llms.txt an official web standard?

No. It's a community-proposed convention, still evolving, and not confirmed as something any major AI engine's ranking or selection process actually reads.

Do I need a developer to publish llms.txt?

No. It's a plain text file. Anyone who can upload a file to a website's root directory (the same place robots.txt lives) can publish one directly from the template above.

How often should I update my hotel's llms.txt?

Whenever the summarized details change meaningfully, a renovated amenity, a new rate structure, updated contact details. It doesn't need a recurring schedule the way review management or AI visibility tracking does.

Related reading
AI Crawler Readiness: How to Make Your Hotel Website Readable by ChatGPT and Gemini →Hotel Schema Markup: A Practical JSON-LD Guide for Hotel Websites →What Is GEO for Hotels? A Practical Guide to Generative Engine Optimization →

See where your own hotel stands.

The calculator takes thirty seconds, costs nothing, and shows its math.

Calculate my revenue at stakeSee all guides →