llms.txt: what the file does and does not do for AI visibility

WebRankInfo published on 3 November 2026 a piece on llms.txt, a proposed file intended to help website publishers interact better with AI systems such as ChatGPT and Gemini. The report makes a point of correcting common assumptions: according to its author, the file is misunderstood by many, and the article sets out what it is really for, why it does nothing today, and what stance to take.

What the proposal actually covers

The report frames llms.txt as a proposal, not an adopted standard. It does not specify which AI providers read such a file, on what schedule, or with what effect on answers or citations. Those details are absent from the announcement as reported, and no adoption figures are given.

That gap matters. A file format becomes useful when the systems that crawl and answer questions actually consume it. Until that is documented and observable, a publisher publishing llms.txt is making a bet on future behaviour rather than acting on a current signal. The report's position is blunt: as things stand, the file has no effect on how AI assistants treat a site.

Why AI visibility cannot be filed away

Getting cited or surfaced by an AI assistant is not a configuration problem. Assistants build answers from crawled pages, retrieved passages and their own ranking choices, and the criteria behind those choices are not published. General industry context is worth recalling here: the mechanisms that decide which sources get quoted change without notice, and no vendor has committed to a contractual relationship with publishers through a text file.

So the practical question for a team is not whether to add one more file to the root directory. It is whether the content that assistants retrieve is clear, well-structured and attributable. That work already exists in most organisations under technical SEO and editorial quality, and it is measurable: server logs show which crawlers arrive, and brand mentions in assistant answers can be sampled and tracked over time.

One caveat from the report deserves emphasis: a lot of commentary around llms.txt is based on assumptions rather than observed behaviour. Treat any claim of a ranking boost with suspicion until someone can show the mechanism.

What to check this week

A short, concrete checklist is enough for most teams:

• Confirm which AI crawlers reach your site, using server logs and robots rules, before debating extra files.
• Sample a set of prompts relevant to your category in two or three assistants and record whether your brand or pages appear, so you have a baseline.
• Review a handful of key pages for extractable, self-contained answers rather than marketing prose, since retrieval favours passages that stand alone.
• If you publish llms.txt anyway, treat it as documentation for humans and as a low-cost experiment, not as a traffic lever.

The report's conclusion points the same way: understand the file, do not expect results from it today, and keep watching for evidence that providers consume it. For acquisition and monetisation planning, that means AI visibility stays a content and measurement discipline, not a checkbox in the site root.