HONEST EXPLAINER
What is llms.txt — and why it won't get you recommended
It's the most-hyped file in AI SEO: a small text file that supposedly tells ChatGPT about your website. Here's what it actually is, what the companies behind the AI assistants say about it, what a 137,000-site study found — and what genuinely moves whether AI names your business.
Free · No sign-up to check · Under two minutes
What llms.txt actually is
llms.txt is a proposed standard — put forward in September 2024 by Jeremy Howard of Answer.AI — for a Markdown file that sits at the root of your website (yoursite.com/llms.txt) and gives AI systems a curated index of your content: a title, a short summary, and organized links to your important pages. A companion file, llms-full.txt, goes further and inlines the full content. Think of it as a sitemap written for language models instead of crawlers.
The idea is reasonable: AI systems have small context windows and messy HTML is expensive to parse, so a clean text version of "here's what this site is" could help them. The proposal took off in developer-documentation circles — some documentation platforms generate it automatically — and the SEO industry picked it up from there, often with claims well ahead of the evidence.
Who actually reads llms.txt today? Effectively nobody.
This is the part the hype skips, so here it is with sources:
- Google doesn't use it and isn't planning to. Google's search team said so directly in mid-2025 — and one of its search advocates compared llms.txt to the keywords meta tag, the classic example of a signal that was ignored because it was self-declared. Google's official guidance for its AI features says the same thing from the other direction: no special files are needed to appear in AI Overviews or AI Mode.
- OpenAI, Anthropic, and Perplexity don't document it either. All three publish guidance for site owners — and it's about robots.txt and crawler user-agents, not llms.txt. None of them lists the file as a signal for what gets cited or named.
- The traffic data agrees. Ahrefs analyzed 137,000 sites with llms.txt files in 2026 and found roughly 97% of the files received zero visits — and across hundreds of millions of AI-crawler requests, hits on /llms.txt were statistically negligible. The bots that decide citations aren't fetching the file.
None of this makes llms.txt a scam — it's a genuine proposal that might get adopted someday. It makes it premature. Today, a file nobody reads can't influence anything, no matter how well you write it.
So should your business bother with one?
Honest verdict: it's harmless, it's cheap, and it currently does nothing — so it belongs at the very bottom of your list, and never instead of the signals that work. If a vendor is charging you for llms.txt setup as an "AI SEO" deliverable, you're paying for a file the assistants don't fetch.
Zarla's position is the same one we apply to our own product. Every published Zarla site ships the things AI systems actually read today — open crawler access, a clean sitemap, structured data, and your services in plain text — automatically, on every publish. We deliberately don't emit llms.txt yet, because the honest data says no major assistant reads it. The day one does, it becomes a one-click addition. That's the right order of operations for your business too: real signals first, speculative files last.
What are the real signals? For a local business, the evidence points the same way every time: your Google Business Profile, your reviews, and a website that answers the customer's question. In our checks, the businesses ChatGPT names carry their Google ratings and review counts right in the answer — the lists are grounded in Google's business data, not in anyone's llms.txt. The full playbook is in our guide to getting your business recommended by ChatGPT.
Forget the file — read the actual answers.
The only reliable way to know where you stand with AI assistants is to ask them. We do it free: the question your customers ask, on ChatGPT and Google's AI, answers shown word for word with the date.
Free to build and preview · Pay only to go live · Cancel any time
Common questions about llms.txt
A proposed standard for a Markdown file at your website's root (yoursite.com/llms.txt) that gives AI systems a curated index of your site — a title, a summary, and organized links to your key pages. A companion variant, llms-full.txt, inlines full page content. It was proposed in September 2024 by Jeremy Howard of Answer.AI, and it's a real, well-intentioned spec — the catch is simply that major AI assistants don't currently read it.
By the major AI systems, no — and that's not our opinion, it's the public record. Google has said it doesn't use the file and compared it to the long-ignored keywords meta tag; OpenAI, Anthropic, and Perplexity document robots.txt and crawler access for site owners, not llms.txt; and Ahrefs' 2026 study of 137,000 sites found about 97% of llms.txt files received zero traffic. Some developer-documentation tools generate and consume it in coding contexts — that's the one niche where it has real, if modest, use.
No — nothing requires it, and skipping it costs you nothing today. Google's official guidance for AI Overviews and AI Mode explicitly says no special files or markup are needed; its AI features are built on the same systems as regular search. The files that do matter remain the boring ones: robots.txt that doesn't block the crawlers you want, and a clean sitemap. Every published Zarla site ships both automatically.
As a five-minute experiment, harmless. As a paid deliverable or a strategy, no — you'd be optimizing a file the deciding systems don't fetch. The worth-it list for AI visibility is short and unglamorous: a complete Google Business Profile, steady recent reviews, a website that states your services and area in plain text, and consistent business details everywhere. Spend there first; revisit llms.txt if and when a major assistant announces support.
If you want one anyway: create a plain-text Markdown file with an H1 title, a one-paragraph summary of your site, then H2 sections containing lists of links to your important pages with one-line descriptions, and upload it to your site root as /llms.txt. The format spec lives at llmstxt.org. Just do it with clear eyes — you're future-proofing for a standard that might get adopted, not improving today's AI visibility.
The signals the assistants demonstrably read: your Google Business Profile (in our checks, ChatGPT's local recommendations arrive with Google ratings and review counts attached), your reviews, and a website that answers the customer's question in plain text. Start by finding out where you stand — Zarla's free AI Visibility Checker asks ChatGPT and Google's AI about your business and shows you the answers word for word — then work the gaps in that order.
Ship the signals AI actually reads — automatically.
Every published Zarla site ships open crawler access, a clean sitemap, structured data, and your services in plain text — no files to hand-write, no vendor invoices. Free to build and preview.
Free to build and preview · Pay only to go live · Cancel any time