The short answer: your law firm can ship an llms.txt file in about an hour, and there is a real case for doing it, but anyone promising it will get you cited in ChatGPT is selling you something. As of June 2026, Perplexity is the only major AI engine that has publicly confirmed it reads llms.txt. Google has said flatly that it does not use the file and has no plans to. Crawler logs tell the same story: across more than 500 million AI bot visits in one recent 90 day window, only 408 requests targeted llms.txt directly. So the honest position is that llms.txt is cheap insurance, not a ranking lever. Build it correctly, point it at your strongest practice area and attorney pages, and move on to the work that actually moves AI citations.
Here is the full picture, what the file is, who reads it, what Google said, and exactly how to build one for a law firm.
What is llms.txt and where did it come from?
llms.txt is a plain text file written in Markdown that lives at the root of your domain, the same way robots.txt and sitemap.xml do. Jeremy Howard, co-founder of Answer.ai and the Fast.ai research lab, proposed the standard in September 2024. The idea is simple. A website’s HTML is full of navigation, scripts, cookie banners, and layout markup that an AI model has to wade through to find the actual content. llms.txt hands the model a curated map instead: a clean list of your most important pages, each with a one line description, so a generative engine can find the signal without parsing the noise.
Think of it as a sitemap written for language models rather than search crawlers. A sitemap.xml tells Google which URLs exist. An llms.txt tells an AI engine which URLs matter and what each one is about. For a law firm, that means you get to point the model directly at your practice area pages, your attorney bios, and your FAQ content, rather than hoping it digs them out of a menu heavy template.
Do AI engines actually read llms.txt?
This is where the marketing and the reality split. The accurate June 2026 answer is mixed, and you need the specifics.
Perplexity is the clear yes. By mid 2025 Perplexity confirmed support, following Anthropic, and it retrieves llms.txt to help prioritize which pages to surface. Since Perplexity is pure retrieval and prints clickable citations next to every claim, a clean llms.txt that highlights your best pages has a plausible, if modest, upside there.
OpenAI is unconfirmed. ChatGPT and SearchGPT have not committed to using llms.txt, though some practitioners report correlated changes in citation patterns after publishing one. That is correlation, not a documented signal, so treat it as a maybe.
Google is a hard no. At Search Central Live in July 2025, Gary Illyes stated plainly that Google does not support llms.txt and is not planning to. John Mueller, in June 2025, said that no AI system currently uses llms.txt and compared it to the old meta keywords tag, the one search engines abandoned over a decade ago because site owners controlled it and gamed it. In December 2025 an llms.txt file briefly appeared in Google’s own developer docs and was removed the same day, with Mueller reiterating there is no official support.
The crawler data backs the skeptics. One analysis of more than 500 million AI bot visits over 90 days found only 408 requests that hit llms.txt directly. That is a rounding error. The files are largely not being fetched in production answer systems today.
So should a law firm bother?
Yes, with the right expectations. The case for shipping one rests on three points.
It is cheap. A correct file for a typical firm takes under an hour and you maintain it once a quarter. There is no recurring cost.
Perplexity already uses it, and Perplexity is the leading indicator. The engine that adopts a signal first is often where you see the earliest movement. Being present costs you nothing and positions you if other engines follow.
The discipline is the real payoff. Writing an llms.txt forces you to decide which ten or fifteen pages on your site actually represent your firm. Most law firm sites have forty pages of thin, templated practice area copy and no clear hierarchy. The act of curating your best pages, writing a tight description for each, and cutting the filler is the same work that improves your AI visibility across every engine, llms.txt or not.
What you should not do is treat llms.txt as a substitute for the moves that genuinely drive AI citations: self-contained answer blocks at the top of every page, LegalService and Attorney schema, a populated Avvo and Martindale presence, FAQ sections, and consistent NAP data across directories. Those are the levers. llms.txt is a nice to have that sits on top of them.
How to build an llms.txt file for a law firm
The format is defined at llmstxt.org and it is strict in a few specific ways. The parser will not recognize the file without the first two elements.
Start with an H1 of your firm name. Follow it immediately with a blockquote, the line beginning with a greater than symbol, that summarizes the firm in one sentence. Both are required. Then organize the rest into four to eight H2 sections, each containing a Markdown bulleted list of links. Each link is a page title, the URL, and a short description under 120 characters that leads with the page’s value, not your brand name, since that description is what the model reads when deciding whether to surface the page.
Here is a working skeleton for a personal injury and family law firm:
# Henderson Law Firm
> Charleston, SC trial firm handling personal injury, car accident, and family law cases since 2004.
## Start Here
- [About Henderson Law Firm](https://example.com/about): Trial record, attorneys, and Charleston office details.
- [Contact and Free Consultation](https://example.com/contact): How to reach the firm and book a case review.
## Practice Areas
- [Car Accident Lawyer Charleston](https://example.com/car-accidents): Fees, process, and what to do after a crash.
- [Personal Injury](https://example.com/personal-injury): Case types handled and how contingency fees work.
- [Family Law and Divorce](https://example.com/family-law): Custody, divorce, and support cases in South Carolina.
## Proof
- [Case Results](https://example.com/results): Verdicts and settlements by case type.
- [Client Reviews](https://example.com/reviews): Verified reviews and outcomes.
## Answers
- [Personal Injury FAQ](https://example.com/faq): Common questions on fees, timelines, and claims.
## Optional
- [Legal Blog](https://example.com/blog): Articles on SC injury and family law.
The Optional section is reserved for lower priority content a model can skip when its context window is tight. Put your blog and resource pages there, not your money pages.
A few rules that trip firms up. Keep every description under 120 characters and lead with what the page does for the reader. Use clean, canonical URLs, no tracking parameters. Limit yourself to your ten to twenty strongest pages, because a bloated file defeats the purpose. There is a companion format called llms-full.txt that inlines the full text of those pages into one file, which is useful for documentation heavy sites but overkill for most law firms.
Where do you put the file and how do you check it?
Place it at the root of your domain so it resolves at yourfirm.com/llms.txt. That is the path the spec defines and the only place a compliant reader will look. If your site runs on WordPress, you can drop the file in the root directory or use a plugin that generates it. On a static or headless setup, add it to your public directory and redeploy.
Once it is live, validate it. Paste the contents into the validator linked from llmstxt.org, which catches the structural problems human eyes miss: trailing whitespace, a missing blockquote, mismatched Markdown, and silently broken links. One broken link or a dropped blockquote and a strict parser ignores the whole file.
One more setting worth knowing. Google’s John Mueller recommended adding a noindex directive to your llms.txt so the file itself does not show up as a result in regular search and clutter your listings. It is a text file meant for machines, not a page you want ranking.
How llms.txt fits the rest of your AI visibility work
llms.txt is the last 5% of an AEO program, not the first 50%. The sequencing that works for law firms looks like this. First, restructure your key pages so each one opens with a 40 to 80 word answer block that directly answers the question in the H1. Second, add LegalService, Attorney, Organization, and FAQPage schema in JSON-LD. Third, populate and align your off-site footprint: Avvo, Martindale-Hubbell, Justia, FindLaw, Google Business Profile, all with identical name, address, and phone. Fourth, build genuine press citations, since independent corroboration is what the trust layer of every AI engine actually rewards. Only after all of that does llms.txt earn its hour, as a clean map laid over a site that is already worth mapping.
Do it in that order and you are optimizing for what AI engines weight today while staying positioned for whatever they adopt next. Do llms.txt first and you have a tidy index pointing at pages that still are not citation worthy.
Frequently asked questions
Does Google use llms.txt in 2026?
No. Google has stated through Gary Illyes and John Mueller that it does not use llms.txt and has no plans to, comparing it to the discredited meta keywords tag. Google Search ranks your pages the normal way regardless of whether the file exists.
Which AI engines actually read llms.txt?
Perplexity has publicly confirmed it reads llms.txt and uses it to prioritize pages. Anthropic has signaled support. OpenAI has not confirmed it, and Google has rejected it. So as of June 2026 the file is a confirmed signal for one major engine.
Will llms.txt get my law firm cited in ChatGPT?
There is no documented evidence that it will. ChatGPT citations are driven by your on-page answer structure, schema, domain trust, and off-site corroboration. llms.txt may help at the margin, but it is not the lever that gets you cited.
How long does it take to create an llms.txt file?
For a typical law firm site, under an hour to draft and validate, then about 15 minutes a quarter to keep links current as you publish or retire pages.
Is llms.txt worth it if so few crawlers fetch it?
For a law firm, yes, because the cost is near zero, Perplexity already uses it, and the curation work improves your site regardless. Just do not let it crowd out the higher impact work of schema, answer blocks, and citations.
What is the difference between llms.txt and llms-full.txt?
llms.txt is a curated index of links with short descriptions. llms-full.txt inlines the entire text of those pages into one file. Most law firms only need llms.txt; the full version suits documentation heavy sites.
If you want to know where your firm actually stands across ChatGPT, Perplexity, and Google AI Mode before you touch a single file, run the numbers with our AI visibility ROI calculator or book a visibility audit and we will show you which moves close the gap fastest.
Tagged