1 What Is llms.txt?
llms.txt is a text file in the root directory of your website (alongside robots.txt and sitemap.xml) that is specifically created for Large Language Models (LLMs) like ChatGPT, Gemini, and Claude. It contains a machine-readable summary of your company, services, and products.
While robots.txt controls what can be crawled, llms.txt is meant to tell AI systems what your company is about and where to find the most important information. That is the idea behind the proposal. Whether a system actually reads the file is a different question (see section 3).
Think of llms.txt as an elevator pitch for AI systems: compact, structured, and with links to the details – provided a system listens to it at all.
2 Structure of an llms.txt File
The file follows a simple Markdown-like structure:
# Company Name > Short description of the company in one sentence. ## About Us More detailed description... ## Services / Products ### Service 1 Description of the service. URL: https://www.example.com/service-1 ### Service 2 Description of the service. URL: https://www.example.com/service-2 ## Contact - Phone: +49 ... - Email: info@example.com - Website: https://www.example.com
3 The Honest Assessment: Who Actually Uses llms.txt?
Since we first published this article in February 2026, the facts have become clearer – and honesty is part of good consulting. In short: none of the major providers has confirmed processing llms.txt. Google has explicitly denied it; OpenAI, Anthropic, and Perplexity have made no commitment, and their crawlers hardly ever request the file.
Google doesn't use llms.txt
Google's official AI guidance (updated June 2026) makes it clear: Google Search, including AI Overviews and AI Mode, requires no additional machine-readable files – llms.txt is not processed and affects rankings neither positively nor negatively.
Industry-wide adoption stands at around 10% of domains – significantly higher among developer tools and documentation platforms (Stripe, Cloudflare, Vercel, Anthropic). Citation studies also show that whether a page gets cited by AI systems depends on passage quality, freshness, and entity signals – not on files in the root directory. In June 2026, Ahrefs measured that 97% of all llms.txt files had received no AI crawler requests at all in May – which underlines the assessment: a supplement, not a foundation.
Whether you set up llms.txt is therefore a question of expectations. The file does no harm and is quick to write. A measurable effect on your AI visibility, however, has not been demonstrated so far – neither for Google nor for ChatGPT, Claude, or Perplexity. We treat it as a side topic. The foundation is crawler access to your pages, content that gives the answer first, complete product data with Schema.org markup, and third-party sources such as review profiles and trade media. How to check all of this is covered in Test AI Visibility.
4 What llms.txt Is Meant to Do for E-Commerce
Online stores typically have thousands of pages: product pages, categories, filter pages, CMS pages. For an AI system, it's difficult to extract the relevant information from this mass. llms.txt is intended as an answer to that – provided a system reads the file:
- Gives a structured overview of your product range
- Links to the most important pages (categories, top products, about us)
- Names your focus areas and how you differ from competitors
- Complements Schema.org with company-level information
5 Best Practices for E-Commerce llms.txt
Keep it brief
Maximum 1-2 pages. AI systems prefer compact information.
Most important first
Core services and USPs in the first paragraph. Like a press article.
Link URLs
Each section should link to the corresponding page.
Update regularly
Incorporate new products, services, or changes promptly.
Mention certifications
Partnerships and certificates strengthen trustworthiness.
6 llms.txt Together with robots.txt and Schema.org
robots.txt and Schema.org are demonstrably relevant; llms.txt is the optional addition:
Controls access: Which AI crawlers can crawl which pages?
Meant to provide context: Who are you, what do you offer, where are the details? Use by the major providers not confirmed.
Provides the data: Structured product information at the page level – for rich results, merchant feeds, and AI agents.