You may have seen ‘add an llms.txt file’ appearing on AEO checklists and wondered whether it is a real signal or an SEO trend looking for a purpose. The honest answer is: it is a real and sensible idea, it is not a magic ranking factor, and whether it is worth your time depends on what kind of site you are running.

What llms.txt Is

llms.txt is a proposed web convention, originally put forward by Jeremy Howard, that suggests placing a plain-text file at the root of your domain — yourdomain.com/llms.txt — containing a structured summary of your site designed for large language models rather than search crawlers.

The idea is straightforward: a robots.txt file tells crawlers what to index; an llms.txt file tells AI systems what your site is about, which pages are most important, and potentially which pages you prefer AI not to train on. It is written in simple Markdown — a brief description of the site, a list of key pages with one-line summaries, and optional sections for what to include or exclude.

Why the Convention Exists

Traditional HTML is rich with metadata that search engines have learned to parse over decades. But AI language models — especially during retrieval-augmented generation — do not necessarily benefit from that infrastructure in the same way. An llms.txt file offers a clean, machine-readable summary that saves an AI system from having to parse navigation menus, cookie banners, and sidebars to understand what a site is actually about.

It also gives site owners a place to signal intent: these are the authoritative pages, this is what I do, this is the entity that owns this domain. Those signals are relevant to answer engine optimization because they reduce ambiguity about your site’s purpose and authority.

Does Any AI Engine Actually Use It?

This is the honest, important caveat: as of now, no major AI search engine has publicly confirmed that llms.txt influences rankings, citations, or retrieval behavior. There is no official statement from Perplexity, OpenAI, or Google saying ‘we read llms.txt and weight it.’ Some AI crawlers may read and use it; most have not published specifics.

That said, several things are true simultaneously: the convention is gaining adoption, Perplexity has expressed interest in structured site metadata, and having a clear factual description of your site available at a standard location is unlikely to hurt anything.

What a Good llms.txt File Looks Like

A minimal, useful llms.txt follows this structure:

  • A one-paragraph description of the site and its purpose — plain factual prose, no marketing language.
  • A list of your most important pages with a one-line description of each — focus on pages that answer common questions about your business or topic.
  • Optional: an ## Optional section listing supplementary pages.
  • Optional: an ## Exclude section for pages you prefer AI systems not use as training data.

Keep it brief and factual. This is not a place for keyword stuffing or promotional copy — if an AI engine reads it expecting a clean summary and finds marketing language, you have made yourself less trustworthy, not more.

llms.txt vs. Existing Signals

I want to be clear about the hierarchy here. A well-written About page, consistent entity signals across the web, quality FAQ content, and genuine off-site mentions are all more impactful for AI citation than an llms.txt file. The file is infrastructure, not strategy.

Think of it the same way you think of a sitemap.xml — it is a good hygiene step that helps AI and search systems navigate your site more accurately. It does not substitute for the content and authority that make you worth citing. The full strategy is covered in entity optimization.

Should You Add One?

If you run a content-heavy authority site, a SaaS product, or a professional services site where you want AI engines to clearly understand your entity and service scope — yes, add one. It takes 30 minutes, it is harmless, and the convention may gain more formal support from major AI engines in the near future. For a five-page brochure site, it is low priority.

At Salterra, I am adding llms.txt to client sites as a default step in new builds and audits — not because the evidence for impact is strong yet, but because the cost is trivial and the convention direction is sensible.

Where This Fits in Your AEO Work

llms.txt is one small piece of technical entity hygiene. Pair it with a complete Google Knowledge Panel, accurate structured data on key pages, and a consistent off-site entity footprint. The AEO overview covers each of these layers and how they work together to build AI citation authority.