What is a llms.txt file?

A llms.txt file is a plain text file placed at the root of a website (accessible at yourdomain.com/llms.txt) that provides AI crawlers and language models with a curated overview of the website's most important content. It follows the llmstxt.org standard proposed in 2024. For a medical practice, it typically includes a brief description of the practice, links to the most important service and procedure pages, FAQ and educational content, location and contact information, and a usage policy describing how AI systems may use the content.

How is llms.txt different from a sitemap.xml file?

A sitemap.xml file is a comprehensive, machine-readable list of every page on a website, intended for search engine crawlers like Googlebot to ensure complete indexing. A llms.txt file is a curated, human-readable (and AI-readable) summary that prioritises a small number of the most important pages, written with descriptive context. Sitemap.xml says 'here is everything that exists.' llms.txt says 'here is what matters most, and here is what it's about.' Both serve different purposes and a well-optimised medical website should have both.

Does my medical practice website actually need a llms.txt file?

As of mid-2026, llms.txt is not yet universally required, but adoption is accelerating quickly among AI search platforms including Perplexity and Claude, which explicitly reference llms.txt during content retrieval. For medical practices specifically — where website content is often deep (many procedure pages, FAQ sections, location pages) and AI systems benefit significantly from prioritisation — llms.txt provides outsized value relative to its implementation cost. The trajectory mirrors XML sitemaps: optional and low-adoption in early years, then standard practice within a few years as AI search becomes a primary discovery channel.

What Exactly Is a llms.txt File — And Does Your Medical Website Actually Need One?

If you've read anything about AI search visibility for medical practices, you've almost certainly encountered the term llms.txt. It's mentioned constantly and explained rarely — usually as a checklist item without context for what it actually is, how it works, or why it matters. This is the explainer. No jargon, no assumed technical background. By the end, you'll understand exactly what this file does and whether your practice needs one.

Where the Idea Came From

In September 2024, a developer named Jeremy Howard proposed a standard at llmstxt.org: a simple convention for websites to provide AI language models with a curated entry point to their content. The proposal recognised a specific problem. AI systems like ChatGPT, Claude, and Perplexity increasingly browse the web to answer questions — but a website's full content (every page, every navigation element, every footer link) is often too much for an AI system to process efficiently when trying to understand what a website is about and find the most relevant information for a specific query.

The llms.txt standard solves this with a simple convention: place a file called llms.txt in the root directory of your website — the same place where robots.txt and sitemap.xml live — written in Markdown format, with a brief description of the website followed by categorised links to the most important pages, each with a short description of what that page contains.

Since the proposal, adoption has grown steadily. Major AI platforms including Anthropic's Claude and Perplexity have indicated their crawlers reference llms.txt files when present. The standard is still young — it is not yet as universal as robots.txt or sitemap.xml — but the trajectory of adoption mirrors how XML sitemaps moved from a niche SEO technique in the early 2000s to a near-universal standard within a few years.

What's Actually Inside a llms.txt File

A llms.txt file is plain text written in Markdown — the same lightweight formatting language used in README files on GitHub and many note-taking apps. It is human-readable (you could open it and understand it) and AI-readable (language models are extremely good at parsing Markdown). For a medical practice, a llms.txt file typically contains: a title and one or two sentence description of the practice — what it is, what specialty, where it's located, and what makes it notable. A short paragraph on usage permissions — whether AI systems may summarise the content, whether attribution is required, and how to request permission for other uses. Then a series of categorised sections with links: "Core Pages" might link to the homepage and the main service/booking pages. "Procedures" might link to each major procedure page with a one-line description of what each covers. "Patient Resources" might link to FAQ pages, patient education content, and blog posts. "Location & Contact" provides address, hours, and contact information in a structured, easy-to-extract format.

The key design principle is curation, not completeness. A medical practice website might have eighty pages — but its llms.txt file might link to only fifteen or twenty, chosen because they represent the practice's most clinically and commercially important content. An AI system encountering this file gets an immediate, structured understanding of what the practice does and where to find the specific information relevant to a patient's query — without needing to crawl and evaluate eighty pages independently.

How AI Systems Actually Use It

When an AI crawler — Perplexity's PerplexityBot, for example — visits a website, it typically checks for a llms.txt file early in its crawl process, the same way it checks robots.txt to understand crawling permissions. If a llms.txt file exists, the AI system uses it as a prioritisation guide: the pages listed in llms.txt are crawled and indexed with higher priority and treated as more representative of what the website is "about" than pages discovered through general site navigation.

For a medical practice with deep website structures — many procedure pages, location pages for multi-site practices, extensive FAQ sections, years of blog content — this prioritisation matters significantly. Without llms.txt, an AI system's understanding of "what does this practice do" is built from whatever pages its crawl happens to discover and weight most heavily, which might not be the pages the practice considers most important. With llms.txt, the practice directly tells the AI system: these are the procedures we want patients to find, this is the FAQ content that answers common questions, this is how to find our locations.

There is a second, more subtle benefit. When an AI system is constructing a response that cites a medical practice — for example, answering "what does [Practice Name] specialise in?" — having a llms.txt file with a clear, well-written description of the practice gives the AI system a high-confidence, authoritative summary to draw from. Without it, the AI must infer the practice's specialisation from scattered content across multiple pages, which produces less consistent and sometimes less accurate descriptions.

What a Good Medical Practice llms.txt Looks Like

The best medical practice llms.txt files we've seen follow a consistent structure. They open with the practice name and a description that includes the specialty, location, and what differentiates the practice — not generic marketing language, but specific factual positioning: "Riverside Dermatology is an independent dermatology practice in Austin, Texas, specialising in medical and cosmetic dermatology including Mohs surgery, acne treatment, and laser resurfacing."

They include a usage policy section — even a short one — that establishes how the practice wants AI systems to engage with its content. This typically permits summarisation and citation with attribution, while reserving rights around full reproduction and AI training use.

They organise links into clear categories with one-line descriptions for each: not just "Services" linking to a single page, but individual links to each major procedure with a description of what each page covers — "Mohs Surgery: skin cancer removal procedure information, recovery timeline, and what to expect" rather than just "Mohs Surgery."

They include a dedicated section for FAQ and patient education content, since this content is frequently the most directly useful for AI systems answering patient questions — and they include clear location and contact information formatted for easy extraction.

So — Does Your Practice Need One?

If your practice has any meaningful website content beyond a single-page brochure site — multiple service pages, an FAQ section, location information, blog content — a llms.txt file provides real value at very low implementation cost. A basic version can be created by a developer in two to three hours using existing site content as source material, and it requires updates only when major new service lines are added.

The honest answer to "do I need one right now, today" depends on how aggressively you're pursuing AI search visibility. If you're not yet thinking about AI search at all, llms.txt is one of many gaps and not uniquely urgent. If you are actively building AI visibility infrastructure — schema markup, AI crawler access, AI-readable content — llms.txt is one of the highest-value, lowest-cost additions to that effort, because it directly improves how efficiently every other piece of that infrastructure gets discovered and used by AI systems.

Iris by AdChoreo includes a llms.txt file as a standard deliverable in every AI visibility engagement, alongside sitemap.xml, robots.txt configuration, ai-index.json, and the full schema stack. If you want to see whether your practice currently has one — and what else is missing from your AI infrastructure — the free agentic readiness audit checks for it directly and scores your practice across all six AI visibility dimensions in 60 seconds.