Insights · October 2, 2026
llms.txt and schema, explained in plain English
Schema is a block of structured facts hidden in your pages that tells Google and AI assistants what your business is, in a format they can't misread. llms.txt is a plain-text briefing file for AI crawlers at yoursite.com/llms.txt. Neither is visible to your customers; both decide how machines describe you.
Updated
What is schema markup?
Schema is a standardized list of facts embedded in a page's code: this is a business, here is its name, phone, and city, these are its services, these are its FAQ answers. Search engines and AI assistants read it directly instead of inferring those facts from your paragraphs — and inference is where machines get things wrong.
When a result shows review stars, FAQ dropdowns, or a complete business card in the sidebar, schema put it there. When an AI assistant states your hours or services correctly, schema is usually why. It has one hard rule: it must match the visible page. Schema that claims what the page doesn't show gets ignored at best.
What is llms.txt?
A plain-text file at yoursite.com/llms.txt that briefs AI systems on your site: who you are, what you offer, which pages matter, how to reach you. Where schema annotates each page, llms.txt summarizes the whole site in one place an AI crawler checks first.
It's a young convention — not every AI system reads it yet. But it costs one text file, the systems that do read it get your story straight from you, and ours is live at saltwaterstudio.xyz/llms.txt if you want to see the format: the same entity sentence we use everywhere, every service with its URL, the portfolio, the contact paths.
Why does consistency matter more than either file?
Machines decide what's true by cross-checking. Your site, your schema, your llms.txt, your Google Business Profile, and the directories all describe your business — and when the descriptions disagree, the machine's confidence drops and it hedges or picks a competitor it's surer about.
This is why we keep one canonical set of facts and repeat it verbatim everywhere: same name, same phone, same city, same sentence describing what the business does. Boring by design. Confidence is built out of repetition without contradiction.
Do I need to do this myself?
No — but someone does, and it should be whoever builds or maintains your site. If you're having a site built, ask the builder two questions: "what schema ships with it?" and "will it be submitted to Google and Bing both?" The answers tell you quickly whether machines were part of the plan or an afterthought. On our builds the answer is in the included list, in writing.
Common questions
Can my customers see schema or llms.txt?
No. Schema sits in the page's code and llms.txt is a plain-text file at yoursite.com/llms.txt. Neither is visible to your customers; both decide how machines describe your business.
Do all AI systems read llms.txt?
Not yet. It's a young convention. But it costs one text file, and the systems that do read it get your story straight from you.
What happens if schema doesn't match the visible page?
It gets ignored at best. Schema has one hard rule: it must match what the page shows.
Why do my site and my listings need to say the same thing?
Machines decide what's true by cross-checking. When your site, schema, llms.txt, Google Business Profile and directories disagree, the machine's confidence drops and it hedges or picks a competitor it's surer about.
What should I ask a web designer about schema?
Two questions: what schema ships with the site, and will it be submitted to Google and Bing both? The answers tell you whether machines were part of the plan or an afterthought.
Related reading and the services that implement all of this — see why businesses miss from ChatGPT answers, the SEO, AEO & GEO service, or the website build.