If you run a barristers’ chambers website, chances are you’re already thinking about SEO, accessibility, and general findability. But there’s a new kind of indexing on the rise — and it doesn’t come from Google.
Large Language Models (LLMs) like ChatGPT, Claude, Gemini and Perplexity are now routinely serving users content that they’ve learned from a wide sweep of the internet. Whether a barrister’s name pops up in a legal chatbot, or your chambers gets recommended for a particular area of practice, these results are increasingly powered by LLMs — not just search engines.
While the LLM landscape is still rapidly evolving, one thing is becoming clear: it’s a good time to decide whether you want your website’s public-facing content included in these models… or not. And to start signalling that decision clearly.
What is llms.txt?
Inspired by the familiar robots.txt file used to guide search engines, llms.txt is an emerging standard that lets you declare your preferences for how your website is used by AI crawlers. You can specify:
-
Which AI companies are allowed or disallowed from accessing your site
-
What parts of your site are off-limits
-
Contact info or policies relating to your content and LLMs
The llms.txt file should sit in the root directory of your website (e.g. yourchambers.com/llms.txt) and follow a simple plain-text format.
Although adoption by major AI companies is still patchy, OpenAI (ChatGPT’s parent), Anthropic, Perplexity and others are beginning to respect this file — and there’s every reason to expect the trend will continue.
Why barrister and chambers should care
Whether you’re enthusiastic about AI or more cautious, it’s important to take control of your content:
-
If you want to be included: You can explicitly allow indexing, and make it easier for LLMs to ingest the right content — e.g. your barristers’ profiles, contact details, practice areas and publications.
-
If you want to stay out: You can disallow indexing entirely, or block specific crawlers. You might choose this for privacy reasons, to retain control over how your content is used, or simply to wait until industry standards are clearer.
Either way, stating your position now gives you a head start and helps avoid default behaviour that may not align with your intent.
Other steps to consider
Besides llms.txt, there are a few other techniques to help guide (or block) LLMs:
1. Meta tags and HTTP headers
Some AI companies are respecting robots meta tags and specific HTTP headers such as:
meta name="robots" content="noai, noimageai"
These can be added site-wide or per-page depending on your CMS setup.
2. Clear copyright statements or AI use policies
It can be useful to publish a page outlining your policy on AI use of your content — whether you allow it, under what conditions, and who to contact. This gives LLM companies a reference point and supports future enforcement.
3. Structured data and clean HTML
If you do want your content to be indexed by LLMs, make sure it’s easy to understand:
-
Use proper HTML heading tags (<h1>, <h2>, etc.)
-
Avoid burying important text in images or scripts
-
Include schema.org structured data for things like people, organisations, and events
These techniques don’t just help search engines — they help language models too.
So, should you opt in or out?
There’s no one-size-fits-all answer. Some chambers may see benefits in allowing their barristers’ expertise to be surfaced in legal AI tools and conversational search engines. Others may prefer to hold off, especially while questions of consent, compensation and accuracy are still evolving.
What matters now is making a conscious decision — and communicating it clearly via your website.
Need help?
We’re monitoring developments in this area closely, and helping our legal-sector clients set up llms.txt, manage meta tags, and shape an AI strategy that fits their goals.
Whether you want to welcome LLMs or shut the door for now, we can help you do it in a way that’s robust, futureproof and aligned with your overall content strategy.
Get in touch if you’d like a hand getting ahead of the curve.
