The llms.txt File Explained: Should You Add One to Your Website?
A new convention is emerging: the llms.txt file that gives AI models direct instructions about your site. Here is what it is, whether it actually works, and whether it is worth implementing.
The llms.txt file is a plain-text convention proposed in late 2024 that works similarly to robots.txt — but instead of telling bots what they cannot crawl, it tells AI models what your most important content is and how to understand your site. Adoption is growing, the format is simple, and it takes under an hour to implement. But does it actually affect AI citations? Here is what we know. Check your overall AI readiness score.
What Is llms.txt?
llms.txt is a Markdown-formatted plain text file placed at yoursite.com/llms.txt. It provides AI language models with a structured, human-readable summary of your site's content and organization.
Unlike robots.txt, which is a technical directive that bots must follow, llms.txt is a suggestion that AI models can choose to use when they crawl and process your content. There is no formal standard — it is a community convention proposed by Jeremy Howard and gaining adoption across the web.
A basic llms.txt looks like this:
# Your Company Name
> A one-line description of what your company does and who it serves.
## Core Documentation
- [Getting Started Guide](https://yoursite.com/docs/getting-started): How to begin using our product
- [API Reference](https://yoursite.com/docs/api): Complete API documentation
- [Pricing](https://yoursite.com/pricing): Plan details and pricing
## Key Blog Posts
- [What Is AEO](https://yoursite.com/blog/what-is-aeo): Introduction to Answer Engine Optimization
- [Schema Markup Guide](https://yoursite.com/blog/schema-guide): Implementation guide for all schema types
## About Us
- [About Page](https://yoursite.com/about): Company background and team
- [Contact](https://yoursite.com/contact): Contact information
Does llms.txt Actually Affect Citations?
The honest answer: early evidence suggests a modest positive effect for some AI platforms, but the data is limited.
What we know:
- →Perplexity has stated they support and read
llms.txtfiles - →ChatGPT has not officially confirmed support
- →Claude (Anthropic) has not confirmed support
- →Google has not confirmed whether Gemini reads it
The mechanism makes logical sense: if an AI crawler reads your llms.txt before crawling the rest of your site, it can prioritize your most important pages and understand your content hierarchy. Whether the major platforms actually implement this in their crawlers is less clear.
The Case For Implementing It
The argument for adding llms.txt is simple: it takes under an hour and costs nothing. Even if only one major AI platform reads it, that is a net positive. Risks are zero.
Additional benefits:
- →Forces you to document your most important pages in one place (useful internally)
- →May be read by smaller AI platforms and enterprise tools that implement the standard
- →As the standard gains adoption, early implementers benefit first
- →Signals AI-readiness as a brand posture
The Case Against Prioritizing It
llms.txt should not be your first AEO investment. The impact ceiling is uncertain, and the foundational signals (Schema markup, bot access, direct answer blocks) have demonstrated, measurable impact. If you have not yet:
- →Fixed bot access in
robots.txt - →Added FAQPage schema to your key pages
- →Rewritten opening paragraphs as direct answers
...then those actions will deliver far higher ROI than llms.txt. Do those first.
How to Write an Effective llms.txt
If you do add it, write it well:
1. Company description line
Write one precise sentence. Not generic ("We help businesses succeed") — specific ("RankAsAnswer analyzes websites for AI citation readiness and generates Schema markup fixes").
2. Curate your links
Only link to your most authoritative, well-optimized pages. The file should point AI models to your best content, not your full sitemap.
3. Use descriptive link text
The link text becomes how AI models understand the page. "About Us" is less useful than "RankAsAnswer company background, founding date, and team credentials."
4. Keep it under 200 lines
Longer files reduce the probability of complete processing. Be selective.
llms.txt vs. Other AEO Priorities
| Action | Implementation Time | Citation Impact | Certainty |
|---|---|---|---|
| Fix robots.txt bot access | 15 minutes | Critical | High |
| Add FAQPage schema | 30 min/page | Very High | High |
| Rewrite direct answer blocks | 30 min/page | High | High |
| Add author schema | 1 hour total | High | High |
| Create llms.txt | 1 hour total | Low-Medium | Low |
Implement llms.txt after the fundamentals are in place — not instead of them. Check your fundamentals score first and come back to this once your AEO score is above 70.