AI search for criminal defence firms

A person who has just been charged with an offence often asks an AI system which lawyer to call before calling anyone. Joe Ashta works on that problem as a GEO / AEO consultant for law firms. This page sets out how the charged person asks, what the answer gives back, why criminal defence is the most exposed practice area, and what a defence firm can do about it.

How a charged person asks

The question usually arrives within hours of the arrest or the release, typed in the words the person would use with a friend: "I was charged with impaired driving in Surrey, who should I call", "my son was arrested last night and court is Monday". Three pressures push it toward ChatGPT, Claude, Gemini or Google's AI rather than a page of links. It is urgent, and one answer beats ten tabs at two in the morning. It is private, and a chat window feels less exposing than a neighbour. And the person does not know the vocabulary, so a plain question that returns a plain answer beats a search term they would have to guess at.

What the answer gives back

The reply is short. It explains the charge, sets out what happens next, then names a handful of firms with a reason attached to each: a defence-only practice, an office in the right city, a lawyer who has run trials on that charge, a law society listing in good standing. Those reasons come from pages the system retrieved while composing the answer, so a firm's public record writes its own sentence. A firm named in the text has a mention; where the answer tells the reader to contact one, that is a recommendation; the source links underneath are citations. Which firms make the list is decided at the moment of asking, by the process on How AI chooses which law firms to name.

Why criminal defence is exposed

Four features of the practice compound each other.

• Referral-heavy. Defence work comes from other lawyers, duty counsel and former clients, all of whom know the firm already. An AI system knows only what it can read, so a reputation held inside the courthouse counts for nothing in the answer.
• Local. The buyer needs someone who appears in a particular courthouse, so the answer turns on whether the system can place the firm in that city.
• Time-critical. The caller is choosing today, often within the hour, so there is no research phase in which an absent firm gets discovered later.
• Low repeat business. Most clients hire a defence lawyer once.

Together those produce a harsh outcome. A defence firm absent from the answer loses the file outright. The caller retains one of the named firms that day and never comes back, and nothing tells them a better-suited firm existed two blocks away.

What the answers reward

The pattern holds across charges and cities. First, a clear entity: one firm name, one office address and one set of lawyer names stated the same way on the site, the directory listings and the law society record, so the system resolves one strong candidate instead of several weak fragments. That is the subject of Entity authority. Second, charge-specific pages that answer the question the buyer typed: impaired driving, domestic assault, drug possession, bail hearings, each with the process, the timeline and the realistic outcomes for the jurisdiction. A general "criminal law" page is retrieved for nothing specific. Third, records the firm does not control: the law society directory, court and news coverage, and bar association publications. Those corroborate what the firm says about itself, and being pulled into the retrieval step at all is Grounding. Being named as the answer is the work called AEO.

The published treatment

Joe Ashta wrote this argument up for the profession in BarTalk, the magazine of the Canadian Bar Association, BC Branch, in April 2026: Beyond Referrals: How Criminal Defence Firms Are Found in the Age of AI Search. It takes the referral question a defence practice has always answered one way and asks what changes when the referrer is a machine that has never met anyone at the firm. Written for practising lawyers, it is listed on the entity record as published writing that a third party edited.

What a defence firm can measure and fix

Start with the questions that bring in files: the charges the firm wants, in the cities where it appears, phrased the way a frightened person would type them. Find out which firms the platforms name for those questions today, and whether yours is among them. Then work the causes of absence in order: name, address and lawyer records that disagree; missing pages for the charges the firm wants; a public record resting only on the firm's own website. Progress is reported as a mention rate, the share of answers naming the firm. Measuring that and fixing what it exposes for such a firm is what Joe Ashta does as a GEO / AEO consultant, and SEO and Google Ads engagements run through NearMe Marketing. The wider picture is on Law firm AI search, and the role itself on GEO consultant for law firms.

Related: AI search for personal injury firms (coming soon), AI search for family law firms and AI search for BigLaw.