The llms.txt guide: what it is, how to write one, and what to put in it
llms.txt is a plain-text guide to your business that lives at yourdomain.com/llms.txt — written for AI crawlers instead of people. It tells the engines what you do, where you are, and what matters, in words you chose. Most businesses don't have one, which is why AI skips them.
Published 2026-07-03 · 8 min read · Grounded AEO
Think of robots.txt: a small file at the root of every website that tells search crawlers where they may go. llms.txt is its younger sibling for the AI era — but instead of rules, it carries a curated summary: your identity, your offer, your key pages, your facts. When an AI crawler lands on your site, it fetches this file first.
What it is not: a magic ranking file. Publishing an llms.txt does not force any engine to recommend you. What it does is remove the main reason engines skip small businesses today — there is nothing on the site written in a form they can reliably read. It's the difference between handing someone a business card and hoping they piece your details together from a parking-lot flyer.
The crawlers behind the major AI engines — the ones deciding who gets named in answers:
→ GPTBot — OpenAI / ChatGPT
→ ClaudeBot — Anthropic / Claude
→ PerplexityBot — Perplexity
→ Google-Extended — Google's AI systems (Gemini, AI Overviews)
One prerequisite: your robots.txt must not block them. Plenty of sites block AI crawlers by accident with a blanket rule — which is one of the checks in the free audit.
- 01
Start with one identity line
First line: business name, what you are, where you operate. "Miller Plumbing is a family-owned plumbing company serving the Denver metro since 2004." An AI reading only this line should already get you right.
- 02
Write the what-you-do paragraph
Three or four sentences covering what you offer, who it is for, and what makes you different — written the way you would tell a neighbor, not a search engine. No keyword stuffing; engines quote clear sentences.
- 03
List your key pages
A short markdown list of the pages that matter — services, pricing, contact, about — each with a one-line description. This tells crawlers where to look instead of leaving them to wander.
- 04
Add a facts section
The verifiable basics in plain lines: address, phone, hours, service area, year founded. These are the facts AI needs to answer "is this business near me and open now?"
- 05
Save it at yourdomain.com/llms.txt
Plain text file, at the root of your site — exactly like robots.txt. If you can upload a file to your website, you can publish this. Check it loads in a browser and you are done.
A fictional Denver plumber, start to finish. Swap in your own facts and this is a publishable file:
# Miller Plumbing Miller Plumbing is a family-owned plumbing company serving the Denver metro since 2004. We handle residential repair, water heaters, and remodels — same-week scheduling, upfront pricing, licensed and insured. ## What we do Residential plumbing repair, water heater replacement and maintenance, kitchen and bath remodel plumbing, and emergency service within our service area. Most jobs are quoted flat-rate before work starts. ## Key pages - Services: https://millerplumbing example .com/services — full service list with flat-rate pricing - Book a visit: https://millerplumbing example .com/book - About: https://millerplumbing example .com/about — the family, the license, the guarantee ## Facts - Address: 1234 W 38th Ave, Denver, CO 80211 - Phone: (303) 555-0142 - Hours: Mon-Fri 7am-6pm, Sat 8am-2pm - Service area: Denver metro (25-mile radius) - Founded: 2004
(URLs are spaced out because the business is fictional — yours would be real links.)
→ Copying your whole website in. It's a guide, not a mirror. Curate.
→ Writing marketing copy instead of facts. “Denver's most trusted!” is unverifiable noise. “Serving Denver since 2004, licensed, flat-rate quotes” is quotable substance.
→ Letting it go stale. Old hours or a dead phone number in your llms.txt actively teaches engines to distrust you.
→ Blocking the crawlers that would read it. Check robots.txt before you bother writing anything.
→ Stopping here. llms.txt is one of several signals — structured facts and a verified record do the rest. The full picture is here.
Does llms.txt replace SEO?+
No. It is a new, separate channel: SEO targets search crawlers ranking links; llms.txt targets AI crawlers building answers. Keep whatever SEO is working — llms.txt is additional ground, and right now it is mostly unclaimed.
How big should the file be?+
One page. A few hundred words is plenty — the point is a curated summary, not a copy of your website. If everything is highlighted, nothing is.
How often should I update it?+
Whenever the facts in it change — hours, services, locations, pricing. A stale llms.txt is worse than none, because it teaches engines your information cannot be trusted. This is exactly what Grounded automates on paid tiers.
Does every AI engine read llms.txt?+
Honestly, not yet — adoption is growing, not universal. But the crawlers behind the major engines already fetch it, the cost of having one is 20 minutes, and the businesses that publish early are training the engines on their own words instead of scraped guesses.
What is llms-full.txt?+
An optional companion file with expanded content — full page text instead of a summary — for crawlers that want depth. Start with llms.txt; add the full version only if you have substantial content worth expanding.
See if the engines can read you today.
The free audit checks your llms.txt, your structured facts, and whether AI crawlers are even allowed in. 30 seconds.