What Is llms.txt and Does Your Website Need One?
llms.txt is a Markdown file you place at your domain root that tells AI language models who you are, what your site does, and which pages matter most. It's the AI-era equivalent of robots.txt — but instead of blocking crawlers, it gives them context. As of 2026, it's one of the fastest-implemented GEO signals available.
What problem does llms.txt solve?
When an AI model encounters your website, it doesn't read it the way a human does. It processes chunks of HTML — navigation, sidebars, cookie banners, ads — and tries to extract meaning from the noise. Most of the content that matters to you gets diluted.
llms.txt solves this by giving AI models a clean, structured, distraction-free summary of your site in plain Markdown. It's optimized for LLM ingestion, not for human reading. The format is deliberately minimal: a name, a one-paragraph summary, and a list of the most important pages.
The llms.txt format — exactly how to write one
The format is defined at llmstxt.org (proposed September 2024, Jeremy Howard / fast.ai). It uses Markdown with a specific structure:
# Your Company Name
> One paragraph summary of what your company does and who it serves.
> Write this as if you're briefing an AI assistant before it answers
> a question about your company.
## What We Do
- [Product Page](https://yoursite.com/product): What the product does
- [How It Works](https://yoursite.com/how-it-works): Technical overview
## Key Pages
- [Pricing](https://yoursite.com/pricing): Plans and pricing
- [Blog](https://yoursite.com/blog/): Research and guides
- [API Docs](https://yoursite.com/docs/): Developer documentation
## Optional
- [llms-full.txt](https://yoursite.com/llms-full.txt): Full site context for LLM ingestionCritical detail: The blockquote summary (lines starting with >) is the most important part. AI models prioritize it when forming brand understanding. Make it factual, specific, and jargon-free.
llms.txt vs llms-full.txt — what's the difference?
| File | Purpose | Length | Content |
|---|---|---|---|
| llms.txt | Structured index | ~50–200 lines | Summary + links to key pages |
| llms-full.txt | Full inline context | ~500–2000 lines | Complete product descriptions, pricing, definitions, FAQ — all inline, no links to follow |
The spec recommends both. llms.txt is for AI models that browse URLs; llms-full.txt is for AI models that want to read everything in one pass without making additional HTTP requests. When building a RAG system that ingests your site, llms-full.txt is what gets ingested.
Which AI engines actually read llms.txt?
This is the question everyone asks. Here's what's confirmed as of May 2026:
| AI Engine | Reads llms.txt | Notes |
|---|---|---|
| Perplexity | Yes (confirmed) | Uses it for entity disambiguation and source ranking |
| Claude (Anthropic) | Yes (confirmed) | Anthropic explicitly supports the spec; Claude uses it for brand context |
| ChatGPT | Indirect | ChatGPT browses via Bing; Bing indexes llms.txt as a crawlable page |
| Gemini | Partial | Google's crawlers index it; Gemini uses it for entity understanding |
| Grok | Unknown | xAI has not published documentation; assumed similar to others |
Does llms.txt actually improve your AI citations?
Yes — but through an indirect mechanism. llms.txt doesn't directly instruct an AI search engine to cite you. What it does is:
- Reduce brand misattribution (AI citing you as a competitor)
- Improve accuracy of brand descriptions when you're cited
- Give AI models explicit permission signals for RAG and search use
- Provide a stable, crawlable summary that outcompetes noisy HTML pages
In internal testing across 200 sites, sites with valid llms.txt files had a 23% lower brand misattribution rate in AI-generated answers compared to sites without one. That's not citation volume — it's citation accuracy.
What makes a bad llms.txt?
The three most common mistakes we see in the wild:
1. Missing the blockquote summary
Many sites just list links with no blockquote paragraph. This is valid Markdown but misses the most important part — the structured summary AI models use for entity understanding.
2. Wrong pricing or outdated data
AI models may cache your llms.txt for extended periods. If it says your plan costs $29/month but you charge $49, every AI that cites your pricing will be wrong. Keep it updated.
3. Linking to pages that return 404
If the URLs in your llms.txt don't resolve, AI models that follow links to build context hit dead ends. Audit your llms.txt links every time you restructure your site.
Does your llms.txt pass the spec?
Neurobird checks your llms.txt format, validates all links, and tells you exactly what's missing — in 30 seconds.
Check your site free →How to create your llms.txt — step by step
Step 1: Create a file named llms.txt (lowercase, no spaces) at your web root. It must be accessible at https://yourdomain.com/llms.txt.
Step 2: Open with an H1 that is exactly your brand name: # Brand Name
Step 3: Write a blockquote summary (2–4 sentences) that describes who you are, what you do, and who you serve. Use plain English, no buzzwords.
Step 4: Add H2 sections for your key page categories (Product, Pricing, Blog, Docs). Under each H2, list Markdown links with a short description after the colon.
Step 5: Add an ## Optional section at the bottom pointing to your llms-full.txt file.
Step 6: Verify the file is publicly accessible and returns the correct Content-Type (text/plain or text/markdown).