- May 12
- 6 min read
Updated: May 20
Billions of people turn to LLMs like Claude, Gemini, and ChatGPT for answers (about 8 billion every month, in fact), which means site owners are looking for ways to make sure these models understand their content. And with the agentic web on the rise—including protocols like Microsoft's NLWeb, which lets AI agents interact with sites via natural language—that need is only growing.
This is where LLMs.txt comes in. LLMs.txt is a file that summarizes a website's key information into a concise, easy-to-digest format that LLMs and agents can use to gather more context about a brand or website. From there, the idea is that search users and agents will have more accurate, more accessible, and more relevant information about a site. (Here's how to adapt your LLMs.txt files for agents.)
What is LLMs.txt, exactly?
LLMs.txt is a simple text file, proposed by data scientist Jeremy Howard, that summarizes a site and tells AI systems how to use the content. LLMs generate text in markdown, and the format of this file is natural for them to read. (See image below as an example.)
“It’s a proposed standard, not a formal one, designed to give websites a clear voice when interacting with LLMs,” says Sviatoslav Pykhnivskyi, machine learning engineer at Wix. “The file provides two things: a concise summary of your site and structured instructions an LLM can easily process, giving you more control over how AI represents your brand.”
This means pointing LLMs toward key pages like product FAQs, policies, or brand guidelines, or telling it to avoid content that you don’t want to appear in AI conversations. For AI agents, it can also be used to expose API endpoints. You can check out other LLMs.txt use cases here.
Adoption of LLMs.txt
LLMs.txt is currently a proposed standard, but so was robots.txt in 1994. This means that adoption has varied across the web with many of the major AI platforms (Nvidia, OpenAI, Perplexity) creating their own LLMs.txt files.
It’s difficult to say how many sites currently use the protocol. But major rollouts from website builders like Wix, plus LLMs.txt generators in tools like XFunnel, have accelerated uptake. (LLMs.txt and other Wix GEO features are included in select Wix plans, at no extra cost.)
In July 2025, Google reps said that Google “does not crawl the LLM.txt files.” But in October 2025, Crystal Carter, Head of AI Search & SEO Communications at Wix Studio, discovered that multiple LLMs.txt files were both crawled and indexed by Google. (Read more about LLMs.txt myths here.)
“What I find fascinating is that we’re seeing LLMs.txt and llms-full.txt files being indexed for large and small websites,” she says, adding that Google is extracting SERP content from LLMs.txt files. “For instance, for Nvidia, the snippet is pulled from halfway down the file, suggesting that they're looking at the entire page.”
In early October 2025, SEO consultant Aimee Jurenka shared a pop-up from a ChatGPT crawl that read: “Accessing text content from llms-full.txt file.”
While additional research needs to be carried out, signs point to wider adoption. Today, Google says,“without this file, agents may spend more time crawling the site to understand its high-level structure and primary content.”
The benefits of LLMs.txt
LLMs.txt is designed to help your site be more discoverable and better understood by LLMs and agents (the core goal of generative engine optimization). By providing a pre-summarized file, you’re essentially telling AI tools what your site is about, which can theoretically lead to better search results, and more accurate conversational outcomes, and sales via agentic commerce.
An LLMs.txt file can potentially guide AI chatbots and agents in the right direction, because instead of letting the AI guess or pull from outdated third-party sources, you’re giving it the official version straight from your site. This helps reduce hallucinations and increases the chances that customers see an accurate, on-message response when they ask an AI about your business or business category.
Raymond Martinez, Vice President of SEO at Archer Education, thought LLMs.txt was especially important for his higher education clients. "I'll be on a call with a partner that has a program ranked number one in the state by many publications. But when you ask ChatGPT about their program, it gives their competitor and says their program is deficient,” Martinez says. “It’s not deficient, but the LLM was probably trained on 2021 data when the program was still being built. So we were like, ‘okay we really need to make sure we have some way to control these outputs.’”
His team created LLMs.txt files that were pinged over 8,000 times in the summer of 2025, according to Martinez’s research. While this is promising, Martinez says it’s still TBD how the file influenced AI outputs. “We're all about finding new ways to optimize for AI search, and that takes experimentation," he says.
The cons of LLMs.txt
A question mark
The biggest con of LLMs.txt is its uncertain future. "We don't know the standard," Pykhnivskyi says. LLMs.txt is a proposal, and while Pykhnivskyi says it’s a valid proposal that has garnered some attention, there’s no guarantee it will become an industry-wide standard.
Still, as mentioned above, there are signs of adoption, and there's no downside to trying it out. (Wix's LLMs.txt files are generated for you, so there's no resource drain on your team.)
Control isn’t guaranteed
LLMs.txt offers a way to signal your preferences (for example, “don’t use product descriptions this way”), but there’s no promise that AI systems will fully honor the content of your file.
You could limit your organic exposure
If poorly implemented, overly restrictive LLMs.txt files might reduce your brand's visibility in AI-generated responses that could otherwise drive discovery.
How to use LLMs.txt on Wix sites
LLMs.txt is currently available to all Wix English-language users and will expand to all languages soon. You don’t need to do anything (except opt-out if you don’t want it for some reason). The system generates the file for you, and it can be found at yourdomain.com/llms.txt. You can also view your LLMs.txt file in your SEO dashboard.
The automatically generated file includes:
your site name
a summary of its content (including blogs and stores)
contact details
a list of products with links
You can edit your site’s LLMs.txt any time in your site's dashboard, but once you edit the file, it will stop updating automatically. If you want to go back, you can reset the file to the default version to resume automatic updates. Read more about LLMs.txt files on Wix.
LLMs.txt best practices
Here are a few things to keep in mind if you’d like to customize your LLMs.txt file.
Follow the format
Whether you're coding from scratch or using an LLMs.txt generator, the recommended structure should be clean and markdown-friendly so LLMs can easily parse it. For example:
Keep it concise
Don’t dump your whole website into the file. The goal is to give AI a curated map of what’s most important. “These files can be pumped up with too many rows, and then they offer no value,” Martinez says.
Test it
Visit https://yoursite.com/llms.txt in your browser. If you can see it, LLMs can, too.
While the future of LLMs.txt is uncertain, Pykhnivskyi believes that staying aware of these new technologies is key. “It’s just one one of many standards that will either take hold or be forgotten, but you need to know what’s emerging and be ready when it’s adopted.”
FAQs
Is LLMs.txt the same as NLWeb?
No. NLWeb is an open protocol from Microsoft that allows websites to answer natural language questions directly, effectively turning your site into a conversational interface. It uses Schema.org markup and RSS feeds to power those conversations.
LLMs.txt is simpler and separate. It's a plain text file that describes your website and its capabilities to AI tools. The two can coexist, and they serve complementary purposes. NLWeb is about answering questions; LLMs.txt is about providing context and instructions. Think of NLWeb as a chatbot layer on top of your content, and LLMs.txt as the briefing document an agent reads before it starts working.
Is LLMs.txt the same as WebMCP?
No. WebMCP is a browser-level standard, currently in development at Google, that allows websites to register specific tools directly inside a web page—things like "book_appointment" or "add_to_cart"—so that a browsing agent can discover and use them in real time. Chrome's Lighthouse agentic browsing audits now check for WebMCP tool registration as part of their scoring.
LLMs.txt works at a higher level. It lives on your server, not in your page code, and it describes your site's structure and capabilities across the whole domain. WebMCP is about what an agent can do on a specific page right now; LLMs.txt is about where to go and what tools exist.
The two are designed to work together as part of a broader agentic protocol ecosystem—alongside MCP servers, A2A frameworks, and commerce protocols—rather than as alternatives to each other. If you're building for the agentic web, both are worth knowing about.
