llms.txt: what it is, who reads it, and how to write one
What llms.txt actually does, which AI tools read it today, and a copy-ready starter template for a local business site, honest expectations included.
llms.txt is having a moment: a plain markdown file at yourdomain.com/llms.txt that tells AI systems what your site is about and where the important pages live. It costs half an hour, and the pitch is that assistants will describe your business more accurately. Here’s what it actually is, who reads it today, what belongs in it, and how to write one that helps instead of embarrasses.
What llms.txt is
The proposal borrows its logic from robots.txt: one predictable location where machines find guidance about a site. But where robots.txt says what crawlers may not do, llms.txt says what they should read. It’s a short markdown file: a title, a one-paragraph summary, then curated sections of links with one-line descriptions. For a local business that means: who you are, where you are, what you offer, and which pages carry the facts.
Honest status: who actually reads it
The straight answer: llms.txt is an emerging convention, not an accepted standard. Some AI tools and crawlers fetch it, others ignore it completely, and none of the major assistants has committed to honoring it. So calibrate accordingly: a well-made llms.txt is low effort with a possible upside and no downside. That’s a reasonable bet. What it is not: a guaranteed visibility boost, and anyone selling it as one is overpromising. The fundamentals (crawler access, structured data, readable HTML) matter more.
What belongs in it
Write the briefing you’d hand an assistant that has ten seconds to describe your business. Business name and a one-sentence description. City and service area. Your core services, each with a link to its page. Opening hours and contact details. Links to your most answer-rich pages (services, prices, FAQ) rather than every page you have. Curation is the whole point: you’re telling machines where the signal is, not handing them a sitemap.
A starter template
An llms.txt is plain markdown, one line at a time. A skeleton to adapt:
- # Harborview Dental (the site name as a level-1 heading).
- > Family dental practice in Rotterdam. Check-ups, implants and emergency appointments; English and Dutch spoken. (A blockquote with the one-paragraph summary.)
- ## Services (a section heading, then one link per line).
- - Emergency appointments (/services/emergency): same-day slots on weekdays.
- - Implants (/services/implants): consultation, pricing and healing timeline.
- ## Contact (address, phone, email and opening hours as plain text).
Common mistakes
- Dumping your whole sitemap into it: ten curated links beat two hundred unsorted ones.
- Writing marketing copy instead of facts: “passionate about excellence” gives a machine nothing to repeat.
- Letting it go stale: a wrong phone number in llms.txt is worse than having no file at all.
- Doing llms.txt instead of the fundamentals: it complements crawler access and structured data; it doesn’t replace them.
The half-hour verdict
If your website fundamentals are in order, llms.txt is a sensible finishing touch: quick to write, easy to maintain, and positioned to pay off if adoption grows. Write it once, set a reminder to re-read it quarterly, and move on. If you build sites for clients, it’s also an easy value-add most competitors haven’t heard of yet.
Check your site in 10 seconds
The free ScanToPitch AI Visibility Check at scantopitch.com/audit checks for llms.txt along with the rest of your AI visibility setup. Instantly, no signup.
Find your next client this week
Scan a city, see which local businesses have weak websites, and send a pitch that references the real problem.
Join the waitlist →