“Share of model” is the next competitive metric PI firms need to track. Gokcen Karaca, head of digital and design at Pernod Ricard, coined the term after discovering that AI models misrepresented his brands, and launched a systematic effort to monitor and reshape how AI described them. As the HBR article frames it, share of model measures "how often and how favorably brands show up in AI results compared with their competitors." For PI firms, that means what percentage of "find me a personal injury lawyer in [market]" queries surface your firm. Firms that invest in authoritative content, specific review language, strong directory presence, and consistent geographic coverage will have higher share of model. Brand investment builds the evidence AI agents draw from.

How AI describes your firm across the web now functions as a ranking input. Research from Carnegie Mellon, cited in the HBR article, shows that even subtle changes in search wording can significantly alter brand recommendations. The researchers used synonyms to alter basic prompts (such as "Help me choose the best VPN service") and found that simple rewording could increase the likelihood of consumers choosing a brand by as much as 78.3%. The same principle applies to how AI agents represent your firm: The language in your reviews, your attorney bios, your practice area pages, and your directory listings shapes how those agents characterize and recommend you. A review that describes "a car accident case in Dallas, handled start to finish, settled for more than I expected" gives AI something specific to work with. Generic five-star reviews do not.

Audit your share of model this week. Open ChatGPT, Perplexity, and Google's AI Overview. Ask each to recommend a personal injury lawyer in your primary market, then record what surfaces and how AI describes your firm. That is your baseline. Next step, according to HBR: Implement an llms.txt file on your website (a machine-readable file that lets AI agents accurately parse your practice areas, geography, and credentials). The HBR article notes that early adopters, including Cloudflare, HubSpot, and Stripe, already do this, with some firms seeing a 12% uptick in AI-generated traffic within two weeks and a 25% increase in organic traffic.

Lock down your name, address, and phone number across every platform your firm appears on. ChatGPT cross-references your data across Google Business Profile, Bing Places, Apple Business Connect, Avvo, FindLaw, Martindale, Justia, and local directories. A mismatch between your GBP address and your Avvo listing reduces ChatGPT's confidence in your firm. Audit every listing. Fix every discrepancy. One firm, one address, one phone number, everywhere. This is not glamorous. This filter decides whether AI recommends you at all.

Add LocalBusiness schema to your website with service areas named explicitly. Schema markup is structured data written for machines, not humans. It tells AI agents your firm type, geography, practice areas, and hours without requiring them to interpret prose. At minimum, implement LocalBusiness schema with your primary practice area, a service area list that names every city and county your firm takes cases in, and attorney credentials. Firms that give AI a clean structured data layer reduce the risk of misrepresentation in AI-generated recommendations. Firms that skip it leave the characterization to inference.

Build location pages that give ChatGPT something specific to quote. Generic service pages ("we handle car accident cases throughout the state") give AI nothing to work with for a precise location query. A page targeting Broward County personal injury cases should name the local courts your firm litigates in, the insurance carriers most active in that market, typical case timelines, and specific case results where your bar association rules allow. When ChatGPT handles a "near me" query with location sharing enabled, it draws from the most specific and authoritative content it can find for that geography. Build pages for every market you want to own.

Build your review volume now and coach clients on what to write. ChatGPT justifies local recommendations by citing patterns across reviews. A firm with 200 reviews that mention "car accident in Fort Lauderdale," "handled my slip and fall in Broward County," and "settled my case faster than I expected" gives AI a location, a case type, and a quality reference it can use. A firm with 200 generic five-star reviews gives AI nothing. When you send a review request, add one line: "If you have a moment, please mention the type of case we handled and the city or area where it happened." That instruction costs nothing and builds the review profile ChatGPT draws from when recommending a firm to someone two miles from your office.

JPMorgan's entry marks a new phase in legal finance. When JPMorgan's asset management division builds a $2.4 billion fund around mass-tort fee streams, it treats contingency-fee litigation as a legitimate alternative asset class. Stephan Mallenbaum, Seeger Weiss's executive director, characterized the deal as "monetizing legal fees it had already earned to receive payment on an accelerated basis." That framing matters: This is a secondary market for earned, anticipated fee income, not speculative investment in unresolved litigation.

The scale of mass tort fees drew institutional interest. A court-appointed panel recommended Simmons Hanly Conroy receive 11.4% of a $2.14 billion opioid litigation fee fund. That is $244 million for one firm in one mass tort. When firms hold expected receivables at that scale, fee monetization becomes attractive to investors managing billions.

For PI firms, access to capital provides a competitive edge. A firm that can monetize earned fees on an accelerated basis can fund more cases, hire earlier, and absorb the costs of prolonged litigation without cash flow constraints. Over time, firms with access to institutional capital build structural advantages over firms that depend entirely on case settlements to fund operations. For PI firm owners, the only question is whether their firm has the fee volume to access it.

The distinction between generative and agentic AI is the story. Legal firms are using AI. The 69% adoption figure from the 8am Legal Industry Report we covered before confirms that. But Anthropic's data reveals that legal professionals use AI almost exclusively in generative mode: drafting briefs, summarizing depositions, researching case law. They do not deploy agents that run intake sequences, execute multi-step case evaluation pipelines, or trigger follow-up workflows without manual intervention. The rest of the professional services world builds AI that acts. Legal is still asking AI to write.

The autonomy gap compounds over time. Anthropic found that experienced users across all industries shift from approving every AI action to monitoring outcomes, auto-approving 40% or more of agent actions (up from 20% for new users). Industries that adopt agentic workflows earlier build institutional muscle for managing autonomous systems. Industries that wait fall behind on both the tools and the operational knowledge required to deploy those systems safely.

The tools just got dramatically more accessible. Claude Code and Cowork have surged. OpenAI just shipped GPT-5.4 with native computer use for the first time. The system can autonomously navigate software, execute multi-step tasks across applications, and operate through the user interface the same way a human would. The barrier to entry for agentic AI just dropped from "hire a developer to build a custom integration" to "give the model access and a job description."

The risk profile explains the hesitation, but not the inaction. Legal work carries high stakes: client data, filed documents, court deadlines. Anthropic's data shows that the high-risk, high-autonomy quadrant remains sparsely populated across all industries, and 80% of all agent tool calls still carry at least one safeguard. The infrastructure for safe agentic deployment exists. Legal firms are choosing not to use it.

Juries Find Meta and YouTube Liable for Addictive Design; 4,000 Related Cases Follow: An L.A. jury found Meta and Google's YouTube liable for deliberately designing their apps to addict a young user, awarding $6 million in damages. A New Mexico jury separately ordered Meta to pay $375 million for failing to protect children from predators on the platform. Both verdicts treat platform design as a defective product rather than protected content, applying a liability theory that mirrors the Big Tobacco playbook. Moody's counts more than 4,000 similar cases pending against 166 companies.

Boies-Led Team Seeks $147M in Fees After $425M Google Privacy Verdict: A 2025 jury found Google liable for collecting app activity data from users who had disabled tracking settings and awarded $425 million in damages. Law firms Boies Schiller Flexner, Susman Godfrey, and Morgan & Morgan reported spending nearly 50,000 hours on the case. Senior attorneys billed as much as $4,000 per hour. The court scheduled a fee hearing for August 2026.

Missouri Jury Awards $82.3 Million in I-70 Crash: A jury awarded $82.3 million to plaintiff Beth Mayhew, including $11.3 million in compensatory damages, $1 million for loss of consortium, and $70 million in punitive damages after a February 2023 collision. Missouri Lawyers Media ranked it the second-highest verdict of 2025 in the state.

NY Trial Lawyers and Albany Lawmakers Unite to Fight Gov. Kathy Hochul's Auto Insurance Reforms: Andrew Finkelstein, president of the 3,500-member New York State Trial Lawyers Association, said Hochul's proposal would reward insurers for delaying and denying valid claims. The governor's package would bar payouts to anyone more than 50% responsible for a crash and narrow the serious injury threshold for pain and suffering claims. The Democratic-led Assembly and Senate both stripped the reforms from their one-house budget plans. Budget negotiations continue.

Law Bear Pursues Arizona ABS License to Operate Non-Attorney-Owned PI Firm: Arizona's Supreme Court established its Alternative Business Structure program in 2021, permitting non-attorneys to own law firms through a formal licensing process. Law Bear, a Phoenix-based personal injury firm, filed for licensure with managing attorney Jason Ruen pending final bar admission. The firm has not yet received regulatory approval and is not currently accepting clients.

Widow Sues Travelers for $2 Billion Over Bad Faith Refusal to Settle Wrongful Death Case: A Madison County jury awarded $241 million against Prairie Farms after a delivery driver's 2016 death from carbon dioxide from dry ice. Johnson's widow now alleges that Travelers, Prairie Farms' insurer, blocked settlement when Prairie Farms wanted to resolve the case, forcing years of unnecessary litigation. Attorney Patrick Salvi Jr. said his client "would have accepted" an earlier settlement.

Legora built its platform on workflow integration, not AI chat. CEO Max Junestrand drew the distinction at a March conference in Stockholm: "It's amazing that everybody can have their own pocket lawyer in Claude, but we're not solving for the same use case." Legora now serves 800 law firms and legal teams and grew from 40 to 400 employees over the past year. Its round drew more than a dozen investors including Benchmark, Bessemer, General Catalyst, Bain Capital, and Salesforce Ventures. The company plans to open offices in Houston and Chicago and grow to more than 300 U.S. employees by the end of 2026.

Harvey drew Sequoia and Singapore's sovereign wealth fund GIC at a $3 billion valuation jump in three months. Harvey expanded from large firm work into specialized practice areas and is pushing into European markets as Legora moves in the opposite direction toward the U.S. Dealroom shows both companies on nearly identical revenue trajectories.

The only jobs AI can't take are client contact and trying cases, and lawyers need to understand that now. Firms that treat attorney time like a document-processing resource are building for a world that no longer exists. The settlement mill model is real: 2,000 cases, three lawyers, overseas paralegals, AI-generated demands, a client success manager handling 500 cases at a time. That model sets the cost floor for the whole industry. You don't have to run it, but you have to know it exists.

Client contact accountability requires a dashboard, not a reminder. Brian's firm tracks client calls through Solidify. When he ran surprise case reviews, his lawyers were 175 calls behind. The day before scheduled reviews, they got it down to 42. He didn't accept either number: "If client contact is the key to your job security, why were you 175 calls behind?" Without a system that surfaces the backlog in real time, showing up for a case review is theater.

Third-party lead gen has a retention problem the industry rarely discusses openly. Brian's firm once brought in more than 100 PI cases per month from lead gen. Sixty days later, they were down to 22 that held. His read: Vendors sold those leads to multiple firms at once, and clients moved to whoever they found on their own. "What happened?” he asked. “There's no business." He stopped looking for better lead gen and started building a brand strong enough to replace it.

Brand equity is geographic insurance against market invasion. In Brian's view, the long-term play is website brand equity that makes it expensive for any competitor to displace you in your own market. And in the AI search era, that insurance comes from SEO and content presence, not from renting attention you don't own.

For any PI firm thinking about what a technical hire might look like, the directory gives you a real picture of the profile. You can see how these lawyers describe their work, what they know how to build, and what problems they chose to solve. The American Bar Association named this the legal engineer role in a 2025 piece: a professional who combines legal knowledge with systems thinking to automate workflows and build tools a firm actually needs. That title may or may not fit the hire you have in mind. The profile is worth understanding either way.

It is also worth browsing as a source of inspiration for PI's biggest operational problems. Intake qualification, lead management, medical records review, document organization—PI firms spend more time on these workflows than on almost anything else, and most of that time goes to work that should not require a lawyer. None of the tools in the LegalQuants directory targets PI specifically. Seeing what a lawyer in Sydney built to track regulatory data across 195 countries, or how a funds lawyer in Hong Kong automated the most painful parts of contract review, sharpens your thinking about what a PI-specific version of that problem-solving could look like.