The Rainmaker System

How to Get Your Law Firm Recommended by ChatGPT and Other AI Assistants

A growing share of people who need a lawyer no longer start with a Google search. They ask an AI assistant a plain question: "I was in a car accident in Novato. Which personal injury law firm should I call?" The assistant answers with two or three names. Whoever gets named gets the call. Everyone else does not exist for that client.

This is not a prediction. We run these queries against live engines as part of our audits, and the pattern is consistent. When we tested the Marin County personal injury market in July 2026, Perplexity answered the question above by naming three firms, and two of them were not even based in Marin County. Established local firms with decades of history and strong credentials were simply not in the answer.

So the question every managing partner should be asking is: how do these answers get built, and how do we get into them?

How AI assistants actually pick which firms to name

ChatGPT, Perplexity, Claude, and Gemini do not maintain a secret ranking of law firms. When someone asks a "which lawyer should I call" question, the engine does something simpler than most people assume. It pulls from a small pool of sources it already trusts, synthesizes what those sources say, and presents the result as an answer.

In legal, that pool is remarkably consistent. Across the markets we have tested, the citations behind AI answers come from the same handful of places:

• Roundup lists like Expertise.com's "best personal injury lawyers in [county]" pages
• Yelp's algorithmic top-10 for the city or county
• Super Lawyers regional pages
• Legal directories like Justia, FindLaw, and Avvo

In our Marin audit, when Perplexity was asked "Who is the best personal injury lawyer in Marin County, California?", it named one attorney, and the citation behind that name was a Justia directory page. Not the firm's website. Not their reviews. A directory listing.

This is the single most important thing to understand about AI visibility: the engine is not ranking you, it is quoting a pool. If you are in the sources it cites, you can be recommended. If you are not, nothing else about your firm matters to the answer. Your case results, your verdicts, your forty years of history: invisible, because the engine never reads them.

The five moves that actually work

Here is the honest playbook. None of this is secret. All of it is work.

1. Fix your structured data

Structured data (schema markup) is machine-readable labeling in your site's code that tells engines exactly what your firm is: a law firm, at this address, in this practice area, with these attorneys and these reviews. Most law firm websites either have none, or have broken schema installed by a website vendor years ago. In our Marin audit, one firm's schema was malformed and actually referenced the vendor's own domain instead of the firm. Another had a full FAQ page with zero FAQ markup, so engines could not use any of it.

Get LegalService or Attorney schema on your site, make sure it names your firm and your city correctly, and add FAQ, attorney, and review markup where you have the content to back it.

2. Publish an llms.txt file

llms.txt is an emerging convention: a plain-text file at yourdomain.com/llms.txt that gives AI crawlers a clean, structured summary of who you are, what you do, and where. It costs nothing to publish and takes an afternoon to write well. Almost no law firms have one. Of the three Marin firms we audited, none did; one returned a 404 for it. It is not a magic switch, but it is a cheap, direct channel to exactly the systems you are trying to reach.

3. Get into the roundups and directories the engines cite

This is the heaviest lever and the most neglected one. Run the buyer queries for your own market, note which sources the AI cites, and get listed in those specific sources. In practice that means claiming and completing profiles on Justia, FindLaw, Avvo, and Super Lawyers, getting onto the Expertise.com list for your county, and making sure your Yelp presence can compete in the local top-10.

Two details matter more than people expect. First, geography: a directory profile filed under the wrong city is close to worthless for local queries. One firm we audited was based in San Rafael but had its directory profiles filed under San Francisco, which meant Marin-specific answers passed it by. Second, consolidation: one strong profile beats eight weak ones. Another audited firm had more than eight separate FindLaw pages, fragmenting its authority across duplicates.

4. Consolidate your review signal

Engines lean on review platforms as a proxy for trust. The goal is not gaming reviews; it is making sure the reviews you have already earned are concentrated where they count. Pick your primary platforms (Google and Yelp for most firms), route happy clients there consistently, respond to what comes in, and clean up duplicate or orphaned listings that split your review count across multiple pages.

5. Publish content worth citing

AI engines quote pages that answer questions directly. A page titled "What is my Marin County car accident case worth?" that actually answers the question, with FAQ schema behind it, is citable. A homepage that says "Aggressive. Experienced. Trusted." is not. Write the questions your clients actually ask, answer them plainly, mark them up, and keep them current.

The honest part: this is a grind, not a project

Everything above is doable by a motivated firm. We tell prospects this directly. But there are two catches, and pretending otherwise would be selling you something.

First, the setup is bigger than it looks. Done properly, the initial work (audit, schema rebuild, llms.txt, directory claims and corrections, roundup placement) is 30 to 40 hours of specialized effort, and most of it sits in the gap between marketing and web development where neither your marketing person nor your web vendor feels responsible.

Second, and more important: it does not stay done. Engines change their behavior constantly. Roundup lists get re-ranked. Competitors get added to the pool. The firms that stay visible are the ones re-running the buyer queries every month, checking what moved, and fixing what slipped. In our experience, that monthly discipline is what firms cannot sustain internally. The first month is exciting. The ninth month is a chore nobody owns. That is where visibility quietly erodes.

Frequently asked questions

How long does it take to show up in AI answers?

There is no fixed timeline, because engines refresh their sources on their own schedules. The controllable part is getting into the cited pool: schema and llms.txt can be live in days, directory corrections in weeks, roundup placement can take longer. Once you are in the sources, re-testing monthly tells you when the answers move.

Is this the same as SEO?

It overlaps but is not the same. Traditional SEO fights for position on a ranked results page. AI visibility is about being present in the small set of sources an engine synthesizes into a single answer. There is no "position three" in an AI answer; you are named or you are not. Much of the work (structured data, directories, reviews) helps both, which is why we treat them as one system.

Can I just do this myself?

Yes, and this article is the map. Budget 30 to 40 hours for a proper setup, plus a standing monthly commitment to re-test queries, maintain placements, and adapt as engines change. If your firm has someone who will genuinely own that indefinitely, do it in house. Most firms do not, which is the reason services like ours exist.

Do AI assistants really send law firms clients today?

They send our own business clients today, which is why we built this system. In legal specifically, the volume varies by market, but the direction is one way: more consumer questions are being answered by assistants, and the answers name a small number of firms. Being one of them costs far less before your competitors figure this out than after.