Guide · AI search & GEO

llms.txt: the no-hype guide for your website

The llms.txt file has become a fixture in every marketing meeting. Here's what it really is, how to build one properly and what to expect from it, minus the magic promises.

Screen showing colorful lines of HTML code on a dark background
Short answer

llms.txt is a plain-text file written in Markdown that sits at the root of your site (yoursite.com/llms.txt) and gives large language models a curated summary of your business plus links to your most important pages. It's a convention proposed in 2024, not an official standard, and Google has said it doesn't need it for AI Overviews. It takes little time to create and does no harm, but it's no substitute for SEO, Schema markup or good content.

Key points
  • llms.txt is an open proposal from 2024, not a W3C or Google standard.
  • Google says you don't need any special files to show up in its AI features.
  • Its real value is with tools and agents that do read it, especially for documentation.
  • It should be an honest, curated summary, not a copy of your entire site or a keyword dump.
  • Pair it with a well-configured robots.txt, Schema JSON-LD and pages that answer specific questions.

What llms.txt is and where it came from

In September 2024, a simple proposal went up at llmstxt.org: websites should offer a file called llms.txt at their root, written in Markdown, with a summary of who they are and a curated list of links to their key pages. The idea is that a language model, which has a limited context window, shouldn't have to guess what matters among menus, banners, scripts and cookie notices.

Think of it as your business's introduction card for a machine that reads fast: name, what you do, who you serve, where, and the five or ten pages that explain you best. While robots.txt says what can be crawled and sitemap.xml lists every URL, llms.txt tries to answer a different question: what should an AI read first to understand you?

The proposal was designed with technical documentation in mind: programming libraries, APIs and manuals, where a coding assistant needs to find the right page quickly. From there it spread to company websites, online stores and local businesses, sometimes with more enthusiasm than evidence.

The spec has kept evolving, and there's a variant, llms-full.txt, that some projects use to pack the full contents of their documentation into a single file. A small business almost never needs it: the basic llms.txt is enough.

What llms.txt is not (and what no one should sell you)

Before you build one, reset your expectations. Here's what we know for certain as of late 2026:

  • It's not an official standard. It's a community-proposed convention. No body governs it, and each AI company decides whether to read it.
  • Google doesn't use it for AI Overviews or AI Mode. Its documentation for AI features says you don't need new machine-readable files, AI text files or special markup; your page just needs to be indexed and eligible to show with a snippet. More detail in our AI Overviews guide.
  • It doesn't guarantee citations. Nobody can promise you that ChatGPT, Perplexity, Gemini or Claude will mention you. Answers depend on the question, on each system's index and on how useful and trustworthy your content is compared with everyone else's.
  • It doesn't replace robots.txt. If you block OAI-SearchBot or PerplexityBot, your llms.txt won't open the door for them.

So why bother? Because it's cheap, because some agents, code editors and AI tools do check it when someone asks them about a site, and because writing it forces you to get your message straight. That clarity usually improves your pages too.

The right structure, step by step

The proposal defines a specific order. Only the title is required, but a useful file includes all of this:

  1. An H1 title with the name of the business or project (a line that starts with a single # sign).
  2. A blockquote (a line that starts with >) with a two- or three-sentence summary: what you do, who you serve and where.
  3. Optional paragraphs or lists with details that help explain the business: hours, service area, what you don't do.
  4. H2 sections (lines starting with ##) that group links. Each link is a Markdown list item: the name in square brackets, the URL in parentheses and, if you like, a colon followed by a short note.
  5. A section called "Optional" for secondary material an agent can skip when it's short on space.

Here's how the top of the file might look for a fictional auto shop in Torreón:

# Ibarra Brothers Auto Repair

> General auto repair and computer diagnostics in Torreón, Coahuila. Serving private vehicles and fleets across La Laguna since 2009.

## Services

- [Tune-ups and diagnostics](https://example.com/tune-up): what's included, turnaround and warranty

- [Brakes and suspension](https://example.com/brakes): common symptoms and inspection

## Contact

- [Location and hours](https://example.com/contact): Blvd. Independencia, Monday through Saturday

## Optional

- [Blog](https://example.com/blog): maintenance tips

If you'd rather not write it by hand, our llms.txt generator builds the file with the correct structure and also suggests robots.txt rules for AI bots.

What to include and what to leave out

The most common mistake is treating llms.txt like a sitemap or, worse, a place to repeat keywords. A language model doesn't reward repetition; it rewards clarity.

Do includeBetter to leave out
Name, line of business and city in the first line of the summaryGeneric phrases like "we're the market leader"
Service pages that explain what, how long it takes and who it's forLegal pages, cart, login or internal search results
FAQs and guides with specific answersLists of hundreds of URLs with no description
Your real service area (Torreón, Gómez Palacio, Lerdo…)Cities you don't serve, added "just in case"
Verifiable facts: year founded, hours, phone numberPrices that change every month and that you'll forget to update

A practical rule: if a new hire read only that file, could they explain to a customer what you sell and where to go to learn more? If the answer is yes, you're on track.

How to publish it and check that it works

Publishing is the easy part. What matters is that it's accessible and consistent with the rest of your site:

  • Location: upload it to the root so it loads at yoursite.com/llms.txt. On WordPress you can upload it through your host's file manager or use a plugin that generates it; on static sites it goes next to index.html.
  • Encoding: UTF-8, so accented characters and special letters display correctly.
  • Server response: open it in your browser and confirm it loads as text, with no odd redirects or error pages in disguise.
  • Access: make sure your robots.txt or firewall isn't blocking the path. If your site sits behind a service like Cloudflare, confirm its bot protection isn't turning away the assistants you actually want.
  • Consistency: the details (phone, address, hours) must match your site, your Schema and your Google Business Profile exactly. If they don't, you create doubt instead of trust.
  • Maintenance: review it every time you change services or URLs. An llms.txt full of broken links signals neglect.

For a full check, run your domain through our free SEO + GEO audit: it checks whether the file exists, whether your robots.txt lets AI assistants in and whether you have structured data.

Where llms.txt fits in a GEO strategy

Compared with the other pieces, llms.txt has the smallest proven impact. What really moves the needle for showing up in AI answers is the usual stuff, done well:

  1. Make sure you can be read: a robots.txt that doesn't accidentally block AI search bots, and a fast site that serves its text in HTML (not hidden behind heavy JavaScript).
  2. Make sure you can be understood: local business Schema JSON-LD with your name, address, hours and service area.
  3. Make yourself worth citing: pages with direct answers, concrete data and real experience. We explain it in how to write content AI wants to cite.
  4. Make sure you can be verified: reviews, mentions on other sites and signals of experience and trust (E-E-A-T).

llms.txt is the sticky note on top of all that. Without the foundation, the note is useless. For the big picture, start with what GEO is and how to show up in ChatGPT.

A real example from La Laguna: what changed and what didn't

A dental clinic in Gómez Palacio (name withheld) asked us for "the file to show up in ChatGPT." We wrote their llms.txt in an afternoon. The valuable part wasn't the file itself but what we found while writing it: their orthodontics page never said how long a typical treatment takes or whether they treat kids, and their address was different on the website and on Google Maps.

We fixed those pages, made the details consistent and added Schema. A few weeks later, the clinic started seeing visits from AI assistants in Google Analytics (we show you how to measure that in this GA4 guide). Was it the llms.txt? Honestly, we can't isolate it, and you should be wary of anyone who says they can. Most likely, clearer content and consistent details did the work.

The lesson: build your llms.txt, but use it as an excuse to get your site in order, not as a shortcut.

If you'd rather review it together, our SEO for AI (GEO) service covers llms.txt, robots.txt, Schema and rewriting the pages your customers ask about most.

FAQ

Questions about this topic

Does Google use the llms.txt file?
No. Google has stated that you don't need special AI files or extra markup to appear in AI Overviews and AI Mode; your page just needs to be indexed and eligible to be shown with a snippet in Google Search.
Do I need llms.txt to show up in ChatGPT?
No. ChatGPT finds and cites websites through its own search index and its OAI-SearchBot crawler. llms.txt is an optional convention: not having one doesn't exclude you, and having one doesn't guarantee you'll be cited.
Where does the llms.txt file go?
At the root of your domain so it loads at yoursite.com/llms.txt, saved as UTF-8 and written in Markdown: a title with #, a summary in a blockquote with > and ## sections with lists of annotated links.
What's the difference between llms.txt and robots.txt?
robots.txt tells crawlers which parts of the site they may visit, and it's a convention the major bots respect. llms.txt neither grants nor removes permissions: it only offers a curated summary and links so a language model can understand the site faster.
How often should I update my llms.txt?
Whenever your services, main URLs, hours or contact details change, and at least once a quarter. A file with broken links or details that don't match your Google Business Profile erodes trust.
Message us on WhatsApp