The honest take
Do you need llms.txt or “AI schema” to rank in AI?
No. Google has said you don’t need special “AI files”, magic “AI schema”, or an llms.txt to be visible in AI search. Anyone selling those as the secret is selling snake oil. Here’s what actually earns AI citations — and how to spot the tricks.
The llms.txt myth
llms.txt is a proposed file that lists your key pages for language models. It sounds official, which is exactly why it gets sold. But Google has publicly stated it does not use llms.txt, and no major AI crawler currently treats it as a requirement. Shipping one is cheap and harmless — we even publish our own as advisory housekeeping — but it does not move the needle. If it’s presented as the lever, that’s your signal to be sceptical.
The “AI schema” / “AI files” myth
There is no special “AI schema”. There’s just Schema.org structured data — the same markup that has helped search engines for years. It’s genuinely useful: it helps machines parse your business and can earn rich results in Google. But it is hygiene, not magic. Markup alone will not make ChatGPT recommend you.
What schema actually does
Structured data describes your pages in a machine-readable way: who you are, your services, your FAQs. That supports entity clarity and rich results. The right move is to ship it correctly and server-side (so AI crawlers, which don’t run JavaScript, can read it) — and then not oversell it. We treat schema as table stakes, not a headline.
What actually earns AI citations
The things that genuinely move your AI visibility are unglamorous and hard to fake:
- Readable pages — server-side rendered content AI can actually parse.
- Consistent facts — one source of truth about you, everywhere.
- A trusted third-party network — accurate mentions across the directories, forums, reviews and articles AI draws from.
- Genuine reputation — real reviews and earned authority placements.
That’s the work. See how to get cited in ChatGPT for the step-by-step, or what GEO is for the overview.
Why we won’t cut corners (and you shouldn’t let anyone else)
AI engines are increasingly aggressive about filtering manipulation — fake reviews, bot comments, spun placements. Providers who automate posting get down-weighted or filtered, sometimes permanently, taking your sources down with them. We monitor and draft with software, but a human posts everything. It’s slower to start and impossible to undo — which is the point. Read more about how we work.
Frequently asked questions
Do I need an llms.txt file to rank in AI?
No. Google has stated it does not use llms.txt, and no major AI crawler currently treats it as required. Publishing one is cheap and harmless housekeeping, but it is not a ranking or citation lever. If an agency sells llms.txt as the secret to AI visibility, treat that as a red flag.
Do I need special “AI schema” for SEO or GEO?
No. There is no special “AI schema”. Standard Schema.org structured data (Organization, FAQ, Article, etc.) is useful hygiene that helps machines parse your pages and can earn rich results — but it is not a magic switch that makes AI cite you.
Does schema markup help AI cite my website?
Indirectly, a little. Schema makes your content easier for machines to parse and disambiguate, which supports entity clarity. But citations are earned through readable content, consistent facts and a trusted third-party network — not through markup alone. Schema is hygiene, not the cause.
How do I tell if a GEO agency is legit?
Ask two questions. One: “How do you post and place content?” If it’s automated bots, walk away — AI engines filter manipulated sources, often permanently. Two: “Are you selling me llms.txt or AI-schema packages as the secret?” If yes, it’s snake oil. Legit GEO earns authority the way the engines actually reward.
Want the honest version for your business?
A free AI visibility audit — no tricks, no upsell pressure. If you don’t need us, we’ll tell you.
Get your free audit