What is llms.txt and how do you create one?
llms.txt is a plain-text file placed at the root of your website (yourdomain.com/llms.txt) that describes, in markdown, what your site is and which pages matter most, written for AI language models. It is a community proposal, not a standard: Google says it does not support it, and major AI vendors have not committed to reading it. This guide covers what it is, why its priority is low, what matters more, and the format if you still want one.
Last updated: · 7 min read
Why llms.txt exists
The idea was proposed as an open convention in 2024: language models work with limited context windows, and a typical website — menus, cookie banners, scripts — is noisy input. The proposal is to hand the model a clean, curated summary instead: who you are, what you offer, where the authoritative pages live. The reasoning is sound, but support did not follow: Google says it does not support llms.txt, no major AI vendor has committed to reading it, and in published measurements of AI crawler traffic llms.txt requests are a very small share.
What matters more than llms.txt
- Crawler access: make sure robots.txt does not block the AI crawlers you want to be read by.
- Readable content: key facts should be in the page HTML your server returns, not only drawn later by JavaScript.
- Structured data: schema.org markup that matches the visible page.
- Consistent brand information: the same name, offering and facts on your site and across the web.
- Visibility on source sites: being mentioned on the pages AI answers cite.
Honest note: llms.txt is low priority. Google says it does not support it and no major AI vendor has committed to reading it. Adding one is harmless, but it is not a visibility lever today; if time is short, fix the items above first.
The format, if you want to add one
The structure is deliberately simple markdown: one H1 with the brand name, a blockquote summary, then H2 sections containing annotated links.
# Acme Software
> Acme is a SaaS platform for SMB accounting automation.
> Trusted by 4,000+ businesses since 2019. HQ: Istanbul.
## Products
- [Acme Invoice](https://acme.com/invoice): E-invoicing and tracking
- [Acme Payroll](https://acme.com/payroll): Automated payroll
## Key Pages
- [Pricing](https://acme.com/pricing): Current plans and prices
- [FAQ](https://acme.com/faq): Top 20 questions answered
## Contact
- Email: support@acme.comThe proposal also describes a companion file, llms-full.txt, for the expanded version — full docs or long-form content.
llms.txt vs robots.txt vs sitemap.xml
| File | Audience | Purpose |
|---|---|---|
llms.txt | AI language models (proposed) | Optional site summary; not supported by Google, no commitment from major AI vendors |
robots.txt | Crawlers/bots, including AI crawlers | Control what may be accessed (widely respected) |
sitemap.xml | Search engines | Enumerate every indexable URL |
If you still want one: 4 steps
Write the header
Start with a single H1 (# Brand Name) followed by a short blockquote (>) that describes in 1-3 sentences what your business is, who it serves and what makes it distinct. This blockquote is the core of the file — it's the part meant to work as a summary.
List your key pages in sections
Group links under H2 headings (## Products, ## Docs, ## Pricing). Each line: a markdown link plus a one-line description after a colon. Only include pages that genuinely matter — this is a curated map, not a sitemap dump.
Upload to your site root
Save as plain UTF-8 text and serve it at /llms.txt (optionally a longer /llms-full.txt). Verify it opens in a browser without a 404 and isn't blocked by robots.txt.
Keep it in sync
If you publish one, update it when products, prices or positioning change. A summary that contradicts your own pages helps no one; your pages themselves are what AI systems actually read.
Or generate it — from verified data
Hand-written files drift out of date and quietly contradict your site. SOV Tracker's fix layer takes the opposite approach: you verify your business facts once (name, offering, prices, FAQs) in the Brand Facts card, and the platform generates schema.org JSON-LD plus an optional llms.txt file from that single source — regenerated whenever the data changes. Generation is available on Starter and above.
Start with Brand FactsCheck your site readiness for free
Fill in the website field below — alongside a live ChatGPT visibility check, the separate Site Readiness check in your scorecard (a technical diagnostic, not part of SOV) checks AI crawler access in robots.txt, schema.org and meta tags, and notes llms.txt for information only (no score weight).
Engine checked: ChatGPT
Frequently asked questions
Is llms.txt mandatory?
No. It's a community proposal, not a standard. Google has said it does not support llms.txt, and no major AI vendor has committed to reading it. Adding one is harmless, but don't expect it to change how AI systems describe you. Put your effort into crawler access (robots.txt), readable content and consistent brand information first.
Where exactly does the file go?
At the root of your domain: https://yourdomain.com/llms.txt — the same location logic as robots.txt. An optional companion, llms-full.txt, can hold the expanded long version of your content.
Which AI systems actually read llms.txt?
Honest answer: none of the major ones says it does. Google states that it does not support llms.txt, and no major AI vendor has committed to reading it. In published measurements of AI crawler traffic, requests for llms.txt are a very small share; crawlers mostly read the HTML pages directly. Our free scorecard notes whether the file exists, but it carries no score weight.
What's the difference between llms.txt and robots.txt?
They are not equals. robots.txt is a long-established standard that crawlers, including AI crawlers, respect: it decides whether they may read your site at all, so it matters directly. llms.txt is an optional summary file that Google and major AI vendors do not support. If you prioritize, get robots.txt right first; llms.txt is optional.
How do I check if my site has llms.txt?
Open yourdomain.com/llms.txt in a browser — if you get a 404, you don't have one. Or run the free scan on this page with your website filled in: the separate Site Readiness check in the scorecard (a technical diagnostic, not part of SOV) checks AI crawler access in robots.txt and schema.org markup, and also notes whether llms.txt exists (for information only; it carries no score weight).
Related tools & guides
llms.txt is a small part of the picture
AI visibility depends far more on crawler access, readable content and where your brand is cited. See the whole GEO playbook — and measure where you stand today.