The llms.txt File That Could Decide Whether AI Search Reads Your Firm's Site At All
A new plain text file is quietly deciding what AI assistants know about a firm before they ever load a page. Here is what llms.txt actually does, and why almost no accountants or solicitors have one yet.
A new client asks ChatGPT to find a solicitor who handles contested probate in Manchester. Before the model names anyone, something has already decided which firms it considers worth reading in the first place. For a small but growing number of sites, that something is a plain text file called llms.txt, and almost no accounting or law firm has ever heard of it.
What llms.txt Actually Is
llms.txt is a single markdown file published at yourdomain.com slash llms.txt. It was proposed in 2024 as a way for a website to hand an AI model a short, curated summary of what the site contains and which pages matter most, written in plain language rather than left for the model to guess at from navigation menus and cookie banners. A growing number of AI tools now check for it before or alongside crawling the rest of a site.
1. Why This Matters More For A Firm Than For Most Businesses
A plumber's website has maybe twelve pages. A multi partner accounting or law firm can have hundreds, split across practice areas, individual fee earner profiles, old news posts, and a blog nobody has touched in three years. When a model only has a limited window to read a site before forming an answer, it tends to read the homepage, maybe one or two linked pages, and stop. An llms.txt file lets a firm point directly at the five or six pages that actually answer client questions, instead of hoping the model stumbles onto them on its own.
2. What Belongs In The File
- A one paragraph description of the firm, written in plain sentences, not a tagline
- A direct link to each practice area page, each with its own single sentence description
- A link to named specialist pages for your three or four most searched client problems, not a general team page
- Your regulator number and the body that issued it, stated as text in the file itself
- A link to a page that states fee ranges or an initial consultation cost
3. What It Does Not Replace
llms.txt is a pointer, not a substitute for the facts themselves. Pointing a model at a page that still says services tailored to your needs with no fee range, no named specialist, and no regulator number does not help, because the file only earns the model's attention for a few extra seconds. ZENETERA has covered what actually belongs on those pages in earlier posts on fee transparency, named specialists, and regulator numbers. The file gets a model to those pages faster. It does not fix a page that has nothing to quote once it gets there.
| Without an llms.txt file | With one that links to the right pages |
|---|---|
| A model reads whatever pages it happens to reach first, often the homepage and nothing else | A model is pointed straight at practice area, fee, and specialist pages in one pass |
| Old or thin pages compete equally with your best content for a model's limited attention | Your best, most specific pages are named as the ones that matter |
| Every firm with a site looks roughly the same to a model skimming quickly | A firm that states exactly what it wants read stands out by being easier to use |
An llms.txt file does not make a weak page strong. It makes sure a strong page actually gets read.
The Takeaway
Most accounting and law firms are still debating whether AI search matters at all, so almost none have published an llms.txt file yet. The ones that add one this year, and link it to a handful of pages that already state real fees, named specialists, and a regulator number, make it easier for a model to find exactly the proof it needs to recommend them. That small head start compounds every time someone asks an AI assistant to find a firm like yours.