Why would a short Markdown file prompt Google to question its value, while Anthropic publishes it on its official website, Shopify enables it by default for merchants, and more AI infrastructure companies move to support it?
The debate around llms.txt may seem like a minor format dispute. In reality, it reflects a larger shift: how brands are read, understood, and recommended by large language models. As part of Tec-Do’s GEO Expert Insights series, this article examines an emerging signal in the market and what it means for brand visibility in the age of generative engines.
Yesterday, a friend in marketing sent me a screenshot. It quoted Google’s John Mueller as saying that llms.txt was not particularly useful and that no major AI system was meaningfully relying on it. The question followed quickly: “We just added llms.txt to our root directory. Was that wasted work?”
I asked him to open claude.com/llms.txt and then look at Shopify, which has already generated llms.txt files for merchants by default. After a pause, he replied: “So who should we listen to?” That is the real issue.
What Is llms.txt: An Instruction Manual for Models
llms.txt is a Markdown file placed in a website’s root directory and written for large language models. Proposed by Jeremy Howard of Answer.AI in September 2024, it was designed to help models read websites more efficiently.
The reason is simple. Commercial websites are often filled with navigation menus, ads, scripts, and pop-ups, while models have limited context windows. llms.txt gives models a cleaner map of the site: what the website does, which pages matter, and how key content should be understood.
It follows the low-cost tradition of robots.txt, but serves a different purpose. robots.txt tells crawlers where not to go. sitemap.xml tells crawlers where content exists. llms.txt tells models what the site is about and what deserves attention.
The Divide: Google Says No, Claude Uses It
The recent debate intensified because Google and Claude appear to be taking different positions.
Google has been skeptical. Mueller compared llms.txt to the old keywords meta tag, arguing that it is only a website owner’s self-description rather than an objective signal assessed by search engines or AI systems.
But the market is not moving in one direction. Anthropic has published llms.txt on Claude’s official documentation site, offering a structured index of its API documents. Shopify has gone further, applying llms.txt across its developer resources and enabling it by default for merchants.
Other adopters include Cloudflare, Vercel, Stripe, Mintlify, Perplexity, Hugging Face, Cursor, and ElevenLabs. Mintlify has even turned llms.txt generation into a default option for projects hosted on its platform.
The contrast is clear: Google questions its usefulness, while model companies, commerce platforms, and documentation infrastructure providers are already experimenting with it.
Reading llms.txt Through MCP: Standards Win on Adoption Cost
This pattern resembles the rise of MCP, the Model Context Protocol introduced by Anthropic in November 2024. At first, it looked like another company-led standard. OpenAI and Google had not adopted it, and the industry response was cautious.
That changed in 2025. OpenAI added native MCP support to ChatGPT and its Agents SDK. Google DeepMind’s Gemini followed. Cursor, Windsurf, Replit, and Zed soon treated MCP as a default plugin protocol.
The lesson is that standards rarely win by formal vote. They win when adoption is cheap and ecosystem value is high. llms.txt has similar conditions: one Markdown file, minimal deployment cost, and potential value if more models, agents, and retrieval systems decide to read it.
A GEO Perspective: Adopt Early, Watch Closely
From a GEO perspective, llms.txt is a low-cost asset with potential upside. It does not guarantee ranking or recommendation, and Google Search may not reward it in the near term. But Claude, Perplexity, vertical agents, and RAG systems may increasingly use structured website cues.
For brands, the practical approach is straightforward. Start with a minimum viable version covering the homepage, product pages, pricing page, and documentation hub. Add short summaries that clarify positioning and priority content. Tec-Do’s llms.txt generator can produce a usable file in about 15 minutes, while quarterly updates are usually enough for maintenance.
It is also important to separate content strategy from access control. llms.txt is an invitation, not authorization. Content a brand wants models to understand can be highlighted there, while restricted areas should still rely on robots.txt and noai controls.
Conclusion
llms.txt is more than a format debate. It is part of a wider contest over the next gateway of digital visibility.
Google remains skeptical, while Anthropic, Shopify, Cloudflare, Vercel, and others are already moving. That split shows the standard is still early, and the decision window remains open.
For marketing teams, llms.txt is not a gamble. It is a low-risk alignment move. Implementing it creates one more chance for a brand to be read by models. Ignoring it may leave that chance to competitors.
The question is not whether to side with Google or with Anthropic and Shopify. The question is how to secure growth before the new rules are fully written.
About Tec-Do
Founded in 2017, Tec-Do is a leading AI MarTech company delivering result-centric marketing solutions for global business growth. Powered by Tec-Chi multi-modal large language models (MLLMs), the company delivers end-to-end marketing solutions through a suite of AI-native, performance-driven products. These products restructure and autonomize mission-critical marketing processes—including market intelligence, content generation, campaign delivery, and performance optimization—across global media channels. In 2025, Tec-Do served over 100,000 advertisers, representing a diversified customer base that spans e-commerce, gaming, entertainment, and local commerce.
