llms.txt is a proposed plain-text file placed at the root of a website (yoursite.com/llms.txt) that lists the URLs the site owner considers most important for AI models to consume. It is modeled on robots.txt but serves the opposite purpose: where robots.txt restricts crawlers, llms.txt curates content for LLM ingestion. As of mid-2026, no major LLM provider has confirmed using it — implementation is a low-cost hedge, not a proven ranking lever.
llms.txt = a markdown-formatted text file at your site root that tells AI models which pages matter most. Proposed by Jeremy Howard in September 2024. One hour to implement. Not yet confirmed as a ranking signal by any major LLM provider, but cheap insurance.
What llms.txt looks like
The proposed format is a short markdown document with this structure:
- H1 title — your site or organization name
- Blockquote summary — one or two sentences describing what your site offers
- Optional context paragraph — additional framing the model can use
- H2 sections with curated links — typically "Best sources", "Comparison", "Pricing", "Docs", each with markdown links plus short descriptions
- Optional llms-full.txt — a parallel file that publishes the actual concatenated content of the curated URLs
Rankio's own llms.txt is publicly visible at rankio.studio/llm.txt. The file gets updated as new high-priority pages ship.
How llms.txt differs from robots.txt
| Dimension | robots.txt | llms.txt |
|---|---|---|
| Purpose | Restrict what crawlers may access | Curate what AI models should prioritize |
| Effect | Enforced — Google and most crawlers honor it | Informational — no engine has confirmed acting on it |
| Format | Allow/Disallow directives | Markdown with H1, blockquote, H2 sections, curated links |
| Year proposed | 1994 | 2024 (Jeremy Howard) |
| Adoption | Universal | Niche — mostly SaaS sites and technical blogs |
Does llms.txt actually work?
As of mid-2026, the honest answer is: no major LLM provider has publicly confirmed using llms.txt to influence training or retrieval. OpenAI, Anthropic, Google, and Mistral have not stated that llms.txt is a ranking signal. Some industry voices (Peec AI published a notable take in early 2026) have called the file overhyped.
Rankio's position is more measured. llms.txt costs roughly an hour to implement, breaks nothing, and is a low-risk hedge if AI providers move to honor it. But it should not be marketed as a guaranteed visibility lever. The signals that demonstrably move AI citation rates today are structured content, schema.org markup, cross-source brand presence, citation-friendly formatting, and topical depth — not llms.txt.
The right framing: implement llms.txt as part of a broader GEO strategy, not as a standalone tactic, and reset expectations accordingly.
How Rankio handles llms.txt
Rankio's content workflow includes llms.txt updates as part of every page publish — when a new high-priority page ships (a comparison page, a glossary entry, a visibility guide), Rankio's publish protocol adds it to llm.txt under the relevant section. This keeps the file fresh without manual overhead, and ensures that if AI providers move toward honoring llms.txt, Rankio-monitored sites are positioned to benefit.
Frequently asked questions
Get the GEO signals that actually move the needle
Rankio measures your visibility across ChatGPT, Gemini, Claude, Perplexity, and Le Chat — and shows you the content gaps that drive real citation gains, llms.txt included.