There is a quiet new convention for the AI era: a plain text file called llms.txt, placed at the root of a website, that tells large language models how to read it. It will not rescue a weak site. But for law firms competing to be cited by AI assistants, it is a low-effort signal that almost no UK firm has adopted yet — which is exactly why it is worth understanding now.
What llms.txt actually is
llms.txt is a Markdown file at yourfirm.co.uk/llms.txt that gives AI systems a clean, curated map of your most important content. Where robots.txt tells crawlers what they may access, llms.txt tells language models what matters and how it is organised — the key pages, what each one covers, and the canonical version of the facts about your firm. It is a guide written for machines that read for meaning rather than for links.
Why it matters more for law firms
Legal websites are often the worst offenders for the things that confuse an AI reader: dense navigation, near-duplicate practice-area pages across multiple offices, important detail trapped in PDFs, and content systems that bury the substance. A clear llms.txt cuts through that. It lets you state plainly which page is the authoritative source on each practice area, where your regulatory information sits, and who the firm’s named experts are — the entity facts that AI search actually cites.
What belongs in a law firm’s llms.txt
A short summary of the firm — who you are, where you operate, what you do, and your SRA regulation. Then a curated list of the pages that matter, each with a one-line description: the core service and practice-area pages, the people, the contact and regulatory pages, and your strongest genuinely useful content. The point is curation, not volume. You are telling the model what to trust, so it should point at your best, most accurate pages, not at everything you have ever published.
What llms.txt is not
It is not a ranking trick, and it is not a substitute for the real work. A model is free to ignore it, and it will do nothing for a site with thin content, inconsistent entity data or no authority. Treat it as the final tidy layer on top of sound foundations — technical SEO, schema markup and content that earns trust — not as a shortcut around them.
How it fits the wider AI-visibility picture
llms.txt is one small, sensible move inside a larger programme: structured data so machines can parse your pages, consistent entity signals across directories, content that answers real legal questions, and the authority that makes assistants trust you. On its own it is a rounding error. As part of a deliberate AI-visibility strategy, it is one more reason for an assistant to read your firm correctly and cite it with confidence.
If you want to know whether your firm’s website is genuinely legible to AI assistants — llms.txt included — the AI visibility audit for SRA solicitors checks it as standard. Or book an intro call.
This pairs naturally with how UK legal directories shape AI recommendations, the off-site half of the same problem.
Technical AI-readiness is one strand of SEO for solicitors done properly — foundations, content and authority working as one.
