Close the gap between how AI sees the firm and how clients describe the firm. Roach uses the EV brand Rivian as a reference: low human brand awareness, high AI brand awareness, because Rivian built content for GEO but less well for human recall. Run the inverse check. Pull the AI answers about the firm from the AI Search Tip below. Compare them against how the firm's last 20 retained clients described why they chose the firm. The delta is the brief for the next quarter of brand work.

Engineer campaigns and content for earned media that LLMs train on. Roach names Reddit, YouTube, and Meta as known sources of LLM training data. For PI firms, the practical translation is podcast appearances that get transcribed, legal-press coverage of verdicts and settlements, and earned commentary in trade outlets. A firm that earns one credible, specific mention in a source LLMs read becomes one citation closer to surfacing in an AI answer. A firm that publishes only on its own website is invisible to that training pipeline.

Pick the category entry points the firm should own. Roach points out that conversational AI queries are longer and more specific than search queries. A person planning a long flight does not search "noise-canceling headphones," they describe a 10-hour flight to Tokyo and ask which headphones fit. The same pattern hits PI. A firm known only as "personal injury in [city]" returns alongside every competitor. A firm known specifically for trucking accidents, traumatic brain injuries from low-speed collisions, or medical malpractice in obstetrics gives the LLM a narrow context to recall the firm by name. Narrow the category, then dominate it across every surface.

You do not need 10,000 subscribers. You need 200. Payload, a newsletter covering the business of space, generates $1.68 million a year from 20,000 subscribers in an industry of 150,000. Co-founder Ari Lewis told the Newsletter Conference they actively discourage subscriptions. They want only readers who work in the industry. The same discipline applies to a PI referral list: small, deliberate, composed of people who control case flow.

A referral newsletter is a relationship channel you own. In an earlier issue, I drew a line between discovery channels and relationship channels. Discovery (billboards, Google, social) makes your firm visible. A newsletter makes your firm trusted. You cannot have coffee with 200 surgeons every quarter, but you can land in their inbox every week. Not to mention, you own a newsletter. Nobody can take that infrastructure from you.

The content lane is wide open. JAMA publishes 87 pages of peer-reviewed "Law and Medicine" research, but the headlines read like this: "US Food and Drug Administration Sentinel Initiative and Postmarket Drug Safety Surveillance." Nobody treating car accident patients opens it. A PI firm can own this intersection with content a busy provider can use: what makes a medical record litigation-ready, how letters of protection work, which documentation practices affect settlement values.

Pick one professional audience and serve it completely. Medical practitioners are the example here, but the principle applies to any referral source. A firm focused on trucking could publish for fleet safety managers. Construction injury practices could target site supervisors. The discipline is choosing one audience and building content for their questions.

Social media builds discoverability. A newsletter builds direct connection. Referral partners might first notice your firm on a social platform. A newsletter is where they come to trust it. The two channels work together: Social earns attention at scale, the newsletter converts that attention into a relationship no algorithm can disrupt.

The system compounds in every direction. A firm that publishes consistent expert content at the intersection of law and medicine becomes a source journalists contact when a story breaks. We covered digital PR as a growth engine in a prior issue. The newsletter feeds media visibility. Media coverage builds AI search authority. AI search drives case generation. Each layer feeds the next.

ChatGPT now competes on Google Search's turf. Google's auction factors in intent signals, quality scores, competitive pressure, and retargeting history to price each click. ChatGPT has none of that infrastructure yet. Gartner VP analyst Nicole Greene told Digiday the CPC model gives buyers "a consistent basis for measuring OpenAI's performance against the rest of their media mix." For PI firms that run six-figure monthly Google budgets, that comparison will happen fast.

The inventory is contextual, not intent-based. ChatGPT ads appear at the bottom of responses on the free and $8 Go tiers. The platform targets by conversation topic, not by search history, demographics, or retargeting data. Ad agency Adthena notes that click prices on Meta run three to five times below Google Search rates because social users are not in an active purchase mindset. The same discount logic could apply here: A user who asks ChatGPT about back pain after a car accident is closer to research mode than to retain-a-lawyer mode.

OpenAI is building the measurement stack in real time. The company posted a job for its first advertising marketing science leader, a role that would own attribution models, incrementality tests, and media mix modeling. Advertisers currently receive only aggregated performance data (total views and clicks) with no access to user conversations, chat history, or personal identifiers. Until independent measurement vendors and clean-room providers are in place, PI firms will buy reach they cannot fully attribute.

$5 per click won't hold. Legal keywords on Google routinely clear $100 per click in competitive markets. If ChatGPT delivers qualified clicks at $5, every PI firm with a Google budget will test it, and the price will rise. If it doesn't, performance buyers measuring cost per signed case will drop out, and the price will fall.

The annual lawsuit rate has dropped to 1.8%. From 1991 to 2005, 7.4% of physicians faced a lawsuit each year, according to research the AMA cites from Jena et al. (2011). That figure dropped to 5% between 2007 and 2008, then to 2.3% in 2016, and to 1.8% in 2024.

The career ever-sued rate has dropped to 28.7%. Between 2007 and 2008, 42.2% of physicians reported facing at least one career lawsuit. By 2016, that share fell to 34.0%. By 2024, it was 28.7%.

Claims per 100 physicians fell from 68 to 56 between 2016 and 2024. The average career claim count per 100 physicians dropped about 18% over that window.

Surgical specialties carry the highest claim density. General surgeons led every specialty with 177 claims per 100 physicians over the course of their careers. Obstetricians-gynecologists followed at 139 per 100. Surgical specialties overall ran at 115 claims per 100, and 46.5% of physicians in those specialties had faced at least one lawsuit. Emergency medicine (42.0% career ever-sued) and radiology (38.2%) rounded out the high-risk specialty roster.

The Middle Atlantic carries higher long-term risk than other regions. Approximately 38% of physicians in the Middle Atlantic Census Division reported facing at least one career lawsuit, compared to 26.0% in the West North Central Division. The short-term gap was 3.4% versus 1.6%.

Most filed claims never pay out. The AMA cites Medical Professional Liability Association data showing 65% of claims closed between 2016 and 2018 closed without a verdict, so the parties dropped, dismissed, or withdrew them before trial. Of the 6% that reached a trial verdict, defendants won 89%, and 72% of all closed claims went unpaid.

Purdue Pharma Sentenced to Pay $225 Million to Justice Department: U.S. District Judge Madeline Cox Arleo in Newark sentenced OxyContin maker Purdue Pharma on Tuesday to pay the federal government $225 million, ending the yearslong criminal case against the company accused of fueling the opioid epidemic. The court ordered criminal penalties topping $5 billion, but the plea agreement redirects most of those funds into Purdue's separate $7.4 billion bankruptcy settlement, which distributes money to communities, local governments, and victims. Acting Attorney General Todd Blanche said Purdue "put profits over patient health and safety." Chairman Steve Miller apologized during the six-hour hearing and confirmed Purdue will cease to exist on Friday, transferring its operating assets to a new entity called Knoa Pharma.

Georgia Appeals Panel Wary of Tossing Spoliation Sanction in Fatal Crash Case: A Georgia Court of Appeals panel pressed pressure-washing company Sun Brite Services Inc. on Tuesday over its challenge to a sanction for destroying evidence in a wrongful-death suit brought by the family of Rickerson Marcellus, a pedestrian struck and killed by Sun Brite driver Norman Goodrum on October 25, 2021. After Sun Brite received a preservation letter, CEO Jeffrey Robison traded in his cellphone, and the company failed to preserve Slack messages, call logs, and emails. The trial court instructed the jury that the missing evidence may have shown that Goodrum acted within the course of his employment when the crash occurred. Judge J. Wade Padgett asked Sun Brite's counsel: "How is it that you destroy the thing they're looking for and then argue that it wasn't relevant?"

300 Lawyers and Investors Meet on Outside Investment in U.S. Law Firms: Holland & Knight hosted an invite-only Legal Services Investor Summit on the rise of management services organization (MSO) deals, which let outside investors hold stakes in a law firm's non-legal operations without breaching prohibitions on direct ownership of U.S. firms. Holland & Knight partners Trisha Rich and Josh Porte told Reuters they have closed more than 15 MSO deals in the past six months and are working on 100 more, with private equity or venture capital firms involved in over half. Recent personal injury MSO deals include a January arrangement between Uplift Investors and a Louisiana PI firm and a $125 million deal involving an Arizona PI attorney earlier in April.

Federal Judge Reprimands Former U.S. Prosecutor for AI-Generated Court Filing: U.S. Magistrate Judge Robert Numbers in Raleigh, North Carolina, reprimanded former assistant U.S. attorney Rudy Renfer for submitting a brief in a veterans' healthcare benefits case that contained fabricated quotes and false legal citations generated by an AI tool. Renfer first told the court the errors were "clerical in nature" and admitted to using AI only when the judge questioned him directly. He resigned from the U.S. Attorney's Office after a 17-year tenure and 30-year legal career. Numbers wrote that Renfer's lost employment was punishment enough for now, but added that courts "necessarily and unfortunately" will need to move beyond reprimands to deter further AI-related errors.

Anthropic and Freshfields Strike Joint AI Legal Tools Deal: London-based Freshfields, with more than 2,800 attorneys, will work with Anthropic to develop AI applications for legal services, including legal and market research, contract review, document drafting, and automation of internal workflows. The firm will also receive early access to future Anthropic models and products. Anthropic called the collaboration its "most material partnership with a law firm to date." Freshfields had already rolled out Anthropic's Claude chatbot internally and plans to expand to Anthropic's agentic AI platform, Cowork. The companies did not disclose financial terms.

The price drop opens ChatGPT ads to firms below the largest spenders. A $250,000 commitment matched the monthly ad budgets of only the largest PI firms in competitive markets. A $50,000 minimum is accessible to firms running smaller monthly Google budgets. Firms that previously could not justify a six-figure test commitment can now buy in.

The cut reflects the urgency behind OpenAI's ad rollout. OpenAI projects advertising revenue of $102 billion by 2030 and expects to lose $14 billion this year, per the same Digiday report. The faster OpenAI scales advertiser participation, the faster it builds the data and case studies it needs to grow. Lower minimums move that timetable up.

OpenAI gave advertisers their first ads manager last week. Digiday reported that the new tool gives advertisers real-time monitoring of impressions and clicks for the first time, without going back to OpenAI or an agency intermediary. The layout resembles Google Ads.

Brand eliminates the lead gen treadmill. Amanda draws a clear line between firms that buy leads and firms that earn them. Lead gen produces first-time callers who may never remember the firm's name. Brand produces clients who return, refer family, and search the attorney by name when someone they know gets hurt. Her cost of acquisition runs one-third of the Miami market average.

Intake is a brand experience, not an administrative function. Amanda models her intake process on the Ritz-Carlton's five-senses framework: Every touchpoint should make the caller feel something. Her team sends new clients a Cuban coffee maker with two cups. Eight intake specialists staff the Miami headquarters. Every new client receives a call from a case manager within the first week and an attorney call within 30 days. The scripts lead with compassion, not questionnaires.

Track tenders, not demands. A “tender” means the firm recovered the maximum policy amount for the client. Amanda incentivizes that metric above all others. Her team rings a bell for small tenders and a cowbell for six figures. The logic: Demand volume measures activity, but tender rate measures whether the firm worked each case to its full value. If the demand letter only proves damages totaling 50% of the policy, the case is not ready. Go back, instruct the client to treat and document the injury, and send the demand when the evidence supports the full amount.

The client relationship does not end at case close. Amanda's team sends birthday cards with a $2 bill, runs a swag room where clients shop when they pick up their check, and mails Thanksgiving cards to the full past-client database. Amanda treats every resolved case as the start of the next referral. The client who remembers how the firm made them feel tells their cousin, their coworker, their neighbor.

Practice-specific intake. Smith.ai's product page describes separate intake flows by practice area, with PI calls capturing incident type, date, and liability details. Workers' comp, family law, and other matters route through different flows. The company also notes the platform does not comply with HIPAA requirements, so any flow that touches protected medical history needs a different solution.

Legal-specific call routing. Smith.ai's legal feature list includes conflict checks, complex call routing, and separate call paths for court staff, opposing counsel, existing clients, and prospective clients.

Integrations PI firms already use. Smith.ai lists native integrations with Clio, MyCase, Lawmatics, and Calendly, plus Zapier connectivity. Every connected integration is one less manual handoff between the front desk and the case management system.

Per-call billing. Smith.ai prices on a per-call basis rather than per-minute. That model is easy to budget against and worth a column in any vendor comparison.