A growing share of people looking for a lawyer begin by asking a question in plain language: who handles probate near me, what should I do after a car accident, how is custody decided. More and more often the first answer comes from an AI assistant that has read the web on their behalf.
That sounds like a new discipline. Mostly it is the old one with less room for sloppiness.
How an assistant decides what to say
An assistant answers from pages it can retrieve and read cleanly. It favours pages that state plainly who is speaking, answer the question in the first lines and are consistent with what the rest of the web says about that firm. It struggles with pages where the text is buried in scripts, where the same content appears at five addresses, or where the firm's name, address and phone number differ from one page to the next.
Four things that help
Structure. Clear headings, short paragraphs and direct answers. A practice page that opens with what the service is and who it is for is easier to quote than one that opens with the firm's history.
Structured data. Schema markup tells a machine that this is a legal service, at this address, with these attorneys, and that this section is a list of questions and answers. The platform publishes it on every page.
One address per page. Canonical tags, redirects and sitemaps that agree with each other. When a page moves, the old address should forward to the new one for good.
A summary written for machines. An llms.txt file is a short, plain-text description of a site for AI systems: who the firm is, what it does, where its offices are and which pages matter most. The platform generates one for every site and keeps it current.
Content still decides it
None of that helps if the site has nothing worth quoting. The firms that are cited are the ones that have answered real questions specifically: not “we handle all family law matters” but a clear explanation of how parenting time is decided in their state.
That is why the bundle includes a content engine. Three researched articles a week, each built around a question clients actually ask, approved by a lawyer before it is published, adds up to a library that both people and machines can use.
What nobody can promise
No one controls what a search engine ranks or what an assistant quotes, and anyone who promises a position is selling something else. What a firm can control is whether its site is fast, unambiguous, technically clean and steadily useful.
A quick self-check
- Does every practice page answer its main question in the first two sentences?
- Are your firm name, address and phone number identical everywhere?
- Does each page have exactly one address, with old ones redirected?
- Do your FAQs appear as marked-up questions and answers?
- Have you published anything useful this month?
If the answer to most of those is yes, your site is already most of the way to being ready for AI search. If not, those are the places to start.