AI Search

LLMs Txt Guide

I get asked about llms.txt in every workshop now, usually in the same breath as "is this the new robots.txt?" Short answer: no, and anyone selling it that wa

Terry Samuels Terry Samuels
Founder, SEO Spring Training
·Oct 11, 2026 ·7 min read

I get asked about llms.txt in every workshop now, usually in the same breath as “is this the new robots.txt?” Short answer: no, and anyone selling it that way is overpromising. But it’s also not nothing. I’ve added llms.txt to a dozen sites this year, and here’s the honest practitioner read on what it is, what to put in it, and whether it’s worth your afternoon.

The short version

  • llms.txt is a proposed standard — a plain Markdown file at your root that points AI tools to your best, cleanest content
  • It is not an official Google, OpenAI or Anthropic requirement, and no major engine confirms using it for ranking yet
  • Think of it as a curated table of contents for machines, not a traffic cop like robots.txt
  • Low cost, low risk, possible upside — worth doing once your GEO foundations are already in place
  • It does not replace schema, clean structure or authority — those still do the heavy lifting

What llms.txt actually is

llms.txt is a Markdown file you place at yourdomain.com/llms.txt. The idea, proposed in 2024, is to give large language models a concise, curated map of your site: the pages that matter, described in plain language, with links. Where a sitemap lists every URL for crawlers and robots.txt tells crawlers what to avoid, llms.txt tries to answer a different question — “if an AI had limited context and wanted to understand this site, what should it read first?”

The format is deliberately simple. An H1 with your site or product name, a short blockquote summary, then sections of linked Markdown bullets grouped by topic. Some sites also publish an llms-full.txt that inlines the actual content of key pages so a model can ingest it in one pass without crawling. That’s the whole spec. No XML, no special headers, no verification step.

llms.txt vs robots.txt vs sitemap.xml

These three get confused constantly, so let’s separate them cleanly. They solve different problems and you want all three doing their own job.

Three files, three jobs

✓ What each one does

  • robots.txt — tells crawlers which paths they may and may not fetch
  • sitemap.xml — lists every URL you want indexed, for completeness
  • llms.txt — curates your best pages and explains them in plain language for AI tools

✕ What llms.txt is not

  • Not an access-control file — it blocks nothing
  • Not confirmed as a ranking or citation signal by any major engine
  • Not a replacement for schema or a sitemap
  • Not a substitute for actually being crawlable and authoritative

The key difference: robots.txt and sitemap.xml are mature standards every search engine honors. llms.txt is a community proposal that adoption is still catching up to. That gap is exactly why you shouldn’t treat it as a silver bullet — but also why the effort-to-risk ratio is attractive.

What to put in your llms.txt

Resist the urge to dump every page in there. The entire point is curation. I treat it like writing an index for a smart but busy reader who will only click three or four links.

The structure I use

Start with an H1 naming the site, then a one-line blockquote that states what you do and who you serve. Follow with grouped sections — I usually use Core Pages, Guides, and About — each a short list of links with a few words of description. Lead with the pages that define your expertise and the ones you most want associated with your brand entity. If you run a service business, that’s your main service pages and your best explanatory guides, not your privacy policy.

Build your llms.txt in four steps

  1. Pick your anchorsChoose the 10–20 pages that best define your expertise and entity.
  2. Describe each oneWrite a plain, honest sentence per link — what it covers, who it helps.
  3. Group & orderCluster by topic; put your strongest, most defining pages first.
  4. Publish & checkDrop it at /llms.txt, confirm it loads as plain text, keep it current.

Write the descriptions yourself in clear language. This is content a model may read directly, so the same rules that get you cited by AI apply here: be specific, be accurate, and don’t stuff keywords. A description that reads like a human wrote it for another human is exactly what you want.

Does llms.txt actually work?

Here’s where I’ll be straight with you, because the hype is loud. As of late 2026, no major AI engine has publicly confirmed that it reads llms.txt to decide what to cite or recommend. Google has said it doesn’t use it for Search. The engines that genuinely drive answers are still leaning on the open web, their training data, and live retrieval from pages they can crawl.

So why do it at all? Three honest reasons. First, some AI developer tools and smaller crawlers do consume it, and that footprint is growing. Second, writing a good llms.txt forces you to articulate your most important pages and your entity in plain terms — which is useful discipline regardless of who reads the file. Third, the downside is almost zero: a small static file that can’t hurt your rankings or leak anything. Low cost, uncertain but real upside.

Treat llms.txt like an early bet with a tiny ante. You’re not paying much to be at the table if the standard takes off — just don’t mistake the ante for a winning hand.

Where it fits in a real GEO plan

llms.txt is a finishing touch, not a foundation. If your pages aren’t crawlable, your content doesn’t answer questions cleanly, and your brand isn’t corroborated across the web, a curated text file won’t rescue you. The order of operations matters. Get the real signals right first — the ones the generative engine optimization hub walks through — then add llms.txt as a cheap incremental step.

2024Year llms.txt was first proposed
0Major engines confirming it as a ranking signal
~1hrTime to write a solid one for most sites

My sequencing looks like this in practice. First, make sure the pages you’d list in llms.txt are genuinely strong and marked up properly — the schema work comes before the text file, not after. Second, confirm those pages are earning mentions and citations on their own. Third, publish llms.txt as a clean summary of that already-solid foundation. If you want to automate the monitoring side, several of the platforms in my AI SEO tools roundup will flag whether AI crawlers are hitting the file.

Common mistakes I see

The same errors come up again and again, and all of them come from treating llms.txt as more magical than it is.

  • Listing everything. A 200-link llms.txt defeats the purpose. Curate ruthlessly.
  • Keyword-stuffed descriptions. If a model does read it, stuffing reads as low quality. Write for a person.
  • Letting it go stale. A file pointing to pages you’ve deleted or restructured is worse than none. Review it when you do your sitemap.
  • Expecting ranking lifts. There’s no evidence it moves rankings. Set your expectations to “small insurance policy,” not “growth lever.”
  • Skipping the fundamentals to do this instead. This is the big one. llms.txt is the last 2% after the real GEO work, never a shortcut around it.

Should you add one?

If your foundations are solid, yes — spend the hour. It’s a clean, low-risk way to put a stake in a standard that might matter, and the act of curating your defining pages is worth something on its own. If your foundations aren’t solid, close this tab and go fix those first; llms.txt will still be here when you’re ready.

This is exactly the kind of “what’s hype and what’s real” judgment we spend four days sorting through live, with real sites and honest numbers, at the SEOST Digital Marketing Conference in Chandler, Arizona. If you’d rather learn the AI-search playbook from people actually testing these standards than guess from blog posts, grab a pass and come work through it with us.

Frequently asked questions

Is llms.txt the same as robots.txt?

No. robots.txt controls which paths crawlers may fetch; it’s an access-control file every engine honors. llms.txt is a curated, plain-language map of your best content for AI tools, and it blocks nothing. They do different jobs and you want both.

Do Google and ChatGPT use llms.txt?

As of late 2026, no major engine has confirmed using it to rank or cite. Google has said it doesn’t use it for Search. Some AI developer tools and smaller crawlers do read it, and adoption is growing, which is why it’s worth a low-effort bet rather than a priority.

Where do I put the llms.txt file?

At the root of your domain, as plain Markdown served at yourdomain.com/llms.txt. Some sites also publish an llms-full.txt that inlines the full text of key pages for one-pass ingestion.

Will llms.txt improve my rankings?

There’s no evidence it affects rankings. Treat it as a small insurance policy on an emerging standard, not a growth lever. The real work is still clean structure, strong content and corroborated authority — the core of GEO.

Want to go deeper than a guide?Learn this live at SEO Spring Training — Apr 7–11, 2027 · Chandler, AZ.
Get Your Pass →
Terry Samuels

About the author

Terry Samuels

Founder and host of SEO Spring Training — a practitioner-taught digital marketing conference in Chandler, Arizona. Terry writes from real campaigns, not theory.

Last updated October 2026 · Reviewed by the SEOST team
Learn this live at SEOSTApr 7–11, 2027 · Chandler, AZ · Passes from $897
Get Your Pass →