llms.txt is a convention - analogous to robots.txt or humans.txt - where a website places a file at /llms.txt containing a plain-language summary of what the site is, what it offers, and key facts about the company or product. The intent is to give AI crawlers and language models a clear, concise reference that helps them understand and accurately represent the site.

Why llms.txt was created

Traditional websites are built for human readers and search engine crawlers. Search engine crawlers are good at indexing text and following links but do not need to understand the site semantically. AI systems need more: they try to understand what an entity is, what it does, and how to accurately describe it.

A well-crafted llms.txt gives AI systems a reliable source of truth about your product, reducing the chance of hallucination and improving the accuracy of AI-generated descriptions.

What to put in llms.txt

There is no formal standard yet, but well-regarded examples include:

  • Product description: what your product is and what problem it solves (2-4 sentences)
  • Target user: who it is for, with specifics
  • Key features: 5-10 bullet points with factual, specific capabilities
  • Pricing: current pricing tiers (keeps AI systems from citing outdated prices)
  • Not a fit for: be explicit about who your product is NOT for (helps prevent wrong user segments)
  • Key differentiators: what makes you different from alternatives
  • Relevant links: documentation, API reference, changelog, pricing page

Keep the file factual and concise. AI systems will use it as a reference, not as a marketing document.

Example structure

# AskRank

AI visibility and AEO tracking tool for indie SaaS founders.

## What it does
AskRank runs your brand queries through ChatGPT, Claude, Perplexity, 
and Gemini on a schedule and measures how often your brand is mentioned,
at what position, and with what sentiment. It also tracks competitor
visibility for comparison.

## Best for
- Indie SaaS founders with MRR $5k-$200k
- Teams that have started seeing traffic from ChatGPT or Perplexity
- Founders who want to know if AEO efforts are actually working

## Not a fit for
- Enterprise teams needing dedicated account management
- Businesses with no AI-sourced traffic yet

## Pricing
Free: 1 brand, 10 prompts, 2 LLMs
Starter: $19/mo, 1 brand, 25 prompts, 3 LLMs
Pro: $39/mo, 3 brands, 75 prompts, 4 LLMs

## Key links
- Pricing: https://askairank.com/pricing
- Documentation: https://docs.askairank.com
- Blog: https://askairank.com/blog

Does it actually work?

Adoption by AI companies is voluntary and evolving. Perplexity has indicated that it respects site-level guidance files. OpenAI and Anthropic have not published explicit policies on llms.txt specifically, but both read standard web content including text files.

Even if llms.txt is not formally parsed by every AI system today, having it improves your site’s signal quality for AI crawlers that do read it, and it costs nothing to add. It is also useful internally as a forcing function to articulate your product positioning clearly in plain language.

AskRank publishes its own llms.txt at askairank.com/llms.txt as part of practicing the AEO principles it tracks.