GLOSSARY
llms.txt
llms.txt is a plain-text file published at a website's root, following an emerging informal convention, that gives AI crawlers and language models a concise, structured summary of what a site or company actually does, intended to be easier for a model to parse reliably than scraping full HTML pages.
Why it matters
As AI assistants increasingly answer questions by summarising or citing websites directly, a clear, accurate llms.txt improves the odds that an assistant describes a company correctly rather than working from an incomplete or outdated scrape of its marketing pages. It also complements, rather than replaces, well-structured on-page content and schema markup: it is a convenience layer for a model's own crawler, not a substitute for the underlying page content actually being accurate and well organised in the first place. It also works best when it links back to the canonical, authoritative pages on the site for further detail, rather than trying to contain the entire company description within the file itself.
What goes wrong when it's handled badly
A llms.txt file that goes stale, describing products or positioning the company abandoned months ago, actively misleads any AI assistant relying on it, which is arguably worse than not having one at all since it is presented as an authoritative, current summary. It has to be kept current in the same way any other core company description does. A concrete example is a company publishing an llms.txt file once during a website relaunch and never updating it again, so it still describes a service line the company discontinued two years later, actively steering AI-generated answers toward something that no longer exists. Treating the file as part of the same update process as any other core company page, rather than a one-time technical task, is what keeps it accurate over time. Reviewing it alongside every other major site update, rather than as a separate, easily forgotten task, is what keeps it from quietly going stale.