llms.txt is a plain-text file in Markdown, placed at the root of a website, that gives AI agents a short summary of the site and links to its most useful pages. It's a proposed convention, not a standard, and no major assistant provider we checked documents reading it. For a practice website it's a ten-minute, low-risk addition; it won't change what assistants say about you on its own.
Interest is high, mostly from curiosity:
| Search (monthly) | Volume |
|---|---|
| llms.txt (US) | 5,400 |
| llms.txt (Canada) | 720 |
| llms txt file (US) | 880 |
| what is llms.txt (US) | 480 |
| llms.txt generator (US) | 480 |
| llms.txt example (US) | 170 |
Google Ads Keyword Planner data via DataForSEO, US and Canada, 12-month average to Sep 2026.
What llms.txt is
The proposal describes itself as "a proposal to standardise on using an /llms.txt file to provide information to help agents use a website" (llmstxt.org (opens in a new tab)). The reasoning is practical. A web page wraps its content in menus, scripts and layout, and the proposal notes that "converting it back into clean text is difficult and imprecise." A short Markdown file hands an AI agent the essentials directly.
The format is simple. Only the first line is required:
- An H1 with the name of the site or organization (required).
- A blockquote with a one-paragraph summary.
- Optional paragraphs or lists with more detail.
- H2 sections that list links, each written as
[name](url)with an optional note after a colon. - An optional section titled Optional, for links an agent can skip when it needs less context.
The proposal says the file "would mainly be useful for inference rather than training": it's meant to be read on demand, when an agent is answering a question, rather than collected for model training.
Who proposed it, and who uses it
Jeremy Howard published the proposal on September 3, 2024; the current version, v2, was last modified on August 10, 2026 (llmstxt.org (opens in a new tab)).
Publishing and reading are different things. On the publishing side, the proposal's site says "thousands of sites publish an llms.txt file" and that "the AI labs themselves publish llms.txt files for their own developer docs: OpenAI, Anthropic, and Gemini." It also says coding agents follow them to find documentation, that Chrome's Lighthouse checks for one in its agentic browsing audits, and that Wix generates one for every Wix site. Those last claims come from the proposal's site; we haven't confirmed them with Google or Wix.
On the reading side, here is what the assistant providers themselves document, as of October 22, 2026:
| Provider | What its official documentation says about llms.txt |
|---|---|
| OpenAI | Nothing. Its crawler overview (opens in a new tab) covers OAI-SearchBot, GPTBot, ChatGPT-User and OAI-AdsBot only. |
| Anthropic | Nothing. Its crawler help article (opens in a new tab) covers ClaudeBot, Claude-User and Claude-SearchBot and robots.txt. |
| Perplexity | Nothing. Its crawler documentation (opens in a new tab) covers PerplexityBot and Perplexity-User. |
| Search ignores it. Creating one "will neither harm nor help your site's visibility or rankings in Google Search, as Google Search ignores them" (Google Search Central (opens in a new tab)). |
So the honest summary: some AI agents may read llms.txt, but none of the major consumer assistants a patient uses to find a practice has said it does.
What llms.txt can and can't do
It can:
- Give any agent that looks for it a clean, accurate summary of your practice in one place.
- Point to the pages you most want read: services, locations, insurance, new-patient information.
- Cost almost nothing to create and maintain.
It can't:
- Control access. That's robots.txt. If you block a crawler there, an llms.txt file doesn't override it. See AI crawlers and robots.txt.
- Make an assistant name you. AI answers vary by phrasing, location and day, and no file can make a practice show up in them.
- Replace structured data. Schema markup describes your address, hours and services in a vocabulary search engines document. Google notes that structured data "isn't required for generative AI search," but it still feeds ordinary search features.
- Keep anything private. llms.txt is public. Anyone can read it.
If you want to understand what does influence AI answers, start with generative engine optimization. The short version: accurate, specific pages and consistent listings matter far more than any single file.
An example llms.txt for a medical practice
Illustrative example (fictional practice and domain). Copy the structure and replace every detail with your own.
# Lakeview Family Dental
> General and family dentistry in Madison, Wisconsin. Accepting new patients of all ages. Open Monday to Friday 7:30 am to 5 pm and the first Saturday of each month.
- Address: 1200 Lakeview Ave, Madison, WI 53703
- Phone: (608) 555-0142
- Dentists: Dr. Ana Ruiz, DDS; Dr. Ben Okafor, DMD
- Languages: English, Spanish
- Insurance: Delta Dental, MetLife, Cigna, Wisconsin Medicaid for children
## Services
- [Dental cleanings and exams](https://lakeviewfamilydental.example/cleanings): checkups for adults and children
- [Dental implants](https://lakeviewfamilydental.example/dental-implants): single-tooth implants and implant crowns
- [Emergency dental care](https://lakeviewfamilydental.example/emergency): same-day appointments for pain or broken teeth
## For new patients
- [Insurance and payment](https://lakeviewfamilydental.example/insurance): plans accepted and payment options
- [Contact and directions](https://lakeviewfamilydental.example/contact): phone, map and parking
## Optional
- [Blog](https://lakeviewfamilydental.example/blog): articles on dental care
For a live file to compare against, open https://medfy.ai/llms.txt in your browser. It follows the same structure: a name, a one-line summary, then sections of links.
Never include: patient names or stories, testimonials, appointment details, staff personal contact information, or medical claims you wouldn't publish on your website. Everything in this file is public.
How to add llms.txt to your website
Plan on about ten minutes once you have the facts in hand.
- Gather the facts. Practice name, address, phone and hours exactly as they appear on your Google Business Profile. Mismatches between sources are worse than a missing file.
- Write the file in a plain-text editor, such as Notepad or TextEdit in plain-text mode. Word processors add hidden formatting.
- Follow the format above: H1 name, blockquote summary, key facts, then H2 sections of links to your real service and contact pages.
- Review it for privacy and accuracy. No patient information, no outcome claims, no insurance plans you've dropped.
- Save it as
llms.txtwith UTF-8 encoding. - Upload it to the root of your domain, so it loads at yourdomain.com/llms.txt. On WordPress, use your host's file manager or SFTP. Some hosted website builders don't allow files at the root; ask their support team.
- Check it. Open the address in a browser. You should see plain text, not a "page not found" screen or your home page.
- Set a reminder to update it whenever hours, providers or insurance change.
Common mistakes
- Treating it as a ranking tool. Google says Search ignores it, and the assistants we checked don't document using it.
- Copying a generator's output without reading it. Generators can pull in outdated pages or the wrong summary.
- Letting it go stale. An llms.txt with last year's hours contradicts your website.
- Blocking AI crawlers in robots.txt while publishing llms.txt. The two send opposite signals.
llms.txt vs robots.txt vs sitemap.xml
These files are often confused. They do different jobs:
| File | What it does | Who documents using it | Does a practice need it? |
|---|---|---|---|
| robots.txt | Tells crawlers which URLs they may access | Google, OpenAI, Anthropic and Perplexity, in their crawler docs | Yes, and check its AI crawler rules |
| sitemap.xml | Lists every indexable page on the site | Search engines | Yes |
| Schema markup | Labels facts like address, hours and services inside your pages | Search engines | Yes; try the schema markup generator |
| llms.txt | Summarizes the site and links to key pages for AI agents | The proposal; no major consumer assistant documents reading it | Optional |
Two details from the sources are worth knowing. Google says robots.txt "is not a mechanism for keeping a web page out of Google"; for that you need noindex or a password (Google Search Central (opens in a new tab)). And the llms.txt proposal says a sitemap "isn't a substitute for llms.txt," because a sitemap lists pages without explaining them.
Medfy doesn't sell llms.txt files. Websites Medfy rebuilds include structured data, and Medfy's weekly AI visibility tracking checks whether ChatGPT, Gemini, Claude, Perplexity and Google AI Overviews name the practice. To see whether AI crawlers can read your current site at all, run the free marketing audit, and for the wider picture of tracking tools, see our overview of AI visibility tools.
Frequently asked questions
Does ChatGPT read llms.txt?
OpenAI doesn't say so. Its crawler documentation, checked October 22, 2026, describes OAI-SearchBot, GPTBot and ChatGPT-User and how they treat robots.txt, but says nothing about llms.txt. The proposal's site notes that OpenAI publishes an llms.txt for its own developer docs; publishing one is not the same as reading yours. Treat any claim that ChatGPT uses your file as unconfirmed.
Do I need an llms.txt file?
No. Google says Search ignores llms.txt files and that you don't need AI text files to appear in Google Search, and the assistant providers we checked don't document reading them. It's a low-cost, low-risk addition that may help AI agents that do look for one. Spend your time first on accurate service pages, consistent listings and crawler access.
Where do I put llms.txt on my website?
At the root of your domain, so it loads at yourdomain.com/llms.txt, the same place robots.txt lives. The current version of the proposal also allows files at subpaths, each covering the pages beneath it, but a practice website needs only one. Save it as plain UTF-8 text, upload it through your host or website platform, then open the address in a browser to check it loads.
Is llms.txt the same as robots.txt?
No. robots.txt tells crawlers which URLs they may access, and Google, OpenAI, Anthropic and Perplexity document how their crawlers follow it. llms.txt is a proposed Markdown summary that points AI agents to your most useful pages. The proposal says it can complement robots.txt by providing context for allowed content. One controls access; the other is a guide to what's there.