What Is llms.txt? The 2026 Guide for Business Websites

llms.txt is a plain-text file placed at the root of your website, at yoursite.com/llms.txt, that gives AI systems a clean, structured summary of what your business is, what it sells, what it costs, and where the authoritative pages live. It is written in Markdown, meant to be read by machines rather than customers, and it exists for one reason: when an AI assistant answers a question about your company, you would rather it read a document you wrote than infer your business from a marketing headline and a pricing table it half understood.

Think of it as the AI-era counterpart to robots.txt, with an important difference. robots.txt tells crawlers where they may not go. llms.txt tells them what is true.

Why this exists at all

A modern marketing site is built for humans. Facts are spread across a hero section, an about page, a pricing grid, a footer, and a dozen service pages, wrapped in navigation, animation, and calls to action. A person navigates that easily. A language model working with a limited context window has to reconstruct your business from fragments, and reconstruction is where errors enter: a price from a page you archived, a phone number from a directory listing that was never updated, a service you stopped offering in 2023.

llms.txt collapses all of that into one document that states your facts plainly, in the order that matters, with no markup to wade through. It is the difference between handing someone a spec sheet and asking them to work it out from a brochure.

What actually goes in the file

The convention is loose, which is a feature rather than a gap. A useful llms.txt for a business website generally covers:

• An H1 with your business name, followed by a blockquote summarizing what you do in one or two sentences. This is the part most likely to be quoted verbatim.
• Canonical facts: legal name, founding year, leadership, headquarters and other locations, phone numbers, email, and the canonical website URL. Every one of these must match your website, your Google Business Profile, and your directory listings exactly.
• What you sell, organized by category, with links to the authoritative page for each.
• Pricing, stated in real numbers wherever you are willing to publish them. Concrete figures are the single most quotable thing on most business sites.
• Key pages as a labeled link list, so an assistant can follow up on the right URL rather than guessing.
• Quick answers to the questions buyers actually ask, written as direct question and answer pairs. This section tends to earn the most citations.
• Trust signals: years operating, clients served, guarantees, and anything else verifiable.

The rule that matters more than the format

Every URL in the file must use your canonical hostname. If your site canonicalizes to yoursite.com and your llms.txt links to www.yoursite.com, you have just handed an AI system two different addresses for the same entity and asked it to work out that they are the same company. That is exactly the kind of ambiguity that gets a business left out of an answer. Pick one hostname, verify it against what your site actually canonicalizes to, and use it everywhere. This is the same entity-consistency principle that governs schema markup and Google Business Profile, applied to a text file.

What llms.txt does not do

Being straight about this matters, because there is a fair amount of overpromising in this corner of the industry. No major AI provider has publicly committed to reading llms.txt as a ranking or citation input. It is a community convention that has seen real adoption, not a ratified standard with guaranteed support. Publishing one will not, by itself, get you cited by ChatGPT.

What it does do is cheap and asymmetric. The file takes an afternoon to write, costs nothing to host, cannot hurt your traditional SEO, and gives any system that does read it, whether that is a crawler, an agent browsing on a user's behalf, or a person pasting your URL into an assistant, an unambiguous version of your facts. The downside is a few hours of work. The upside is being correctly represented in an answer you never see. That trade is worth making even at uncertain odds.

llms.txt is not a ranking trick. It is a statement of fact that removes the need for a machine to guess, and machines that guess about your business tend to guess wrong.

Where it fits in a real GEO strategy

llms.txt is the cheapest item on the list, not the most important one. In rough order of impact for most businesses: consistent entity facts across your site, your structured data, and your directory listings; answer-first content that leads with the direct answer instead of burying it under six paragraphs of preamble; FAQ sections whose visible copy matches your FAQPage structured data exactly; a crawler policy that actually permits AI user agents; and then llms.txt as the layer that ties it together. Our 2026 GEO checklist walks the full list, and what GEO and AEO actually mean covers the strategy behind it.

The common failure is doing llms.txt first and treating it as the whole job. A perfect llms.txt on a site that blocks GPTBot in robots.txt, contradicts itself on pricing, and has no structured data is a tidy summary of a confusing entity.

How to write and publish one

• Write it in Markdown and save it as llms.txt at the root of your site, served as plain text at yoursite.com/llms.txt.
• Lead with your business name and a one-sentence summary. Assume the first two lines are all some systems will read.
• Use your canonical hostname in every single URL, with no exceptions.
• State real numbers. Vague ranges are unquotable; specific prices, dates, and counts are what get cited.
• Date the facts section and set a calendar reminder to review it quarterly. A confidently stated stale price is worse than no price at all.
• Cross-check it against your website, your structured data, and your Google Business Profile before publishing. If any two disagree, fix the disagreement rather than the file.

How ChaseDaddy.com handles this

We publish our own at chasedaddy.com/llms.txt, and every site we build ships with one generated from the same source of truth that feeds the site's structured data, so the two cannot drift apart. That is included in the $3,000, $5,000, and $10,000 packages rather than sold as a recurring GEO retainer, alongside the schema markup, answer-first content structure, and AI crawler policy that make the file worth reading in the first place.

If you want to know how an AI assistant currently describes your business, and where it is getting that wrong, book a free 90-minute audit and we will show you exactly what the machines see.