Darrow AI finds the lawsuit hiding in plain sight

The button said no. The software behaved as though someone had said yes. In September 2026, Darrow published research on 404 applications built by seven AI coding agents. Among builds following expert privacy specifications and displaying a working opt-out, 30% still loaded tracking scripts after rejection. The interface offered reassurance. The machinery underneath offered something rather different.

That small betrayal explains Darrow better than a grand speech about artificial intelligence. A possible legal problem leaves traces. Someone must notice them, connect them to an obligation and decide whether they matter. Darrow has made that sequence its business.

The first customer said no

Evyatar Ben Artzi and Elad Spiegelman had studied law and clerked at Israel’s Supreme Court. Their third co-founder, Gila Hayat, brought intelligence and data-science experience. In Hayat’s account, the lawyers’ frustration concerned a system whose traditions made meaningful change difficult. Their useful realization was that identifying new cases could be treated as an intelligence-gathering problem.

The information might exist without arriving in a form a lawyer could use. A public disclosure here, a complaint there: individually, each fragment could look ordinary. Together, they might describe harm. Darrow, founded in 2020 and named for American trial lawyer Clarence Darrow, set out to assemble the picture.

The first commercial response was discouraging. According to Ben Artzi’s account to LawSites, corporate compliance buyers doubted the company could anticipate where plaintiff lawyers would strike. They declined. Darrow turned toward plaintiff firms, whose reason to pay was more immediate: a credible discovery could become their next case.

The distinction is useful for anyone selling an unfamiliar product. An insight can sound speculative to the person asked to prevent a future loss. To the person who can build a business around it, the same insight may look like inventory. Darrow later returned to compliance and insurance after developing its plaintiff-side business.

A lawsuit has a balance sheet

Darrow’s work begins outside the law firm. It examines publicly available information, including regulatory filings, incident reports and litigation patterns. Its legal knowledge layer gives those observations context; AI systems identify and prioritize patterns. The company says experienced legal professionals and specialists review insights before delivery.

The May 2026 portfolio-platform announcement brings discovery, evaluation, portfolio management and interactive intelligence into one workflow. A firm can investigate a potential matter, examine comparables and consider defendant history before committing. Once cases enter the portfolio, the dashboard tracks their stages, settlement estimates and projected economics for the firm. The product page currently describes access as a selected-partner beta.

Underwriting is the less glamorous, more consequential companion. Darrow’s written evaluations cover class size, damages, outcome probabilities and likely timing. They also consider collectibility and the possibility that competing actions will dilute a firm’s share. A plausible allegation and an attractive investment are separate judgments. A lawyer working on contingency has to make both.

“Legal exposure doesn’t announce itself.”

Evya Ben Artzi · May 2026

Subscription and usage fees underpin the business, with coverage and activity affecting scope. Forbes also reported in 2025 that some Darrow-originated matters involved fee participation through Arizona lawyer Don Bivens. Darrow told the publication its partnerships vary by jurisdiction. The commercial arrangement deserves attention because the discovery itself can produce value long after the software has finished its work.

Darrow reports more than $22 billion in surfaced legal exposure. That figure describes identified risk; it is neither company revenue nor damages collected. Its website also reports more than 80 organizations using its intelligence. Those numbers describe the company’s claimed scale, while the practical question for a buyer remains wonderfully unromantic: is this particular matter worth pursuing?

Within legal technology, Darrow occupies the discovery end of the process. LawSites compares parts of its portfolio proposition with EvenUp and Supio, whose focus is more on preparing cases. Darrow’s distinguishing wager is that the valuable work can begin before an attorney has a case to prepare. Conventional investigation remains an alternative, especially where the decisive facts are private.

The button worked. The promise did not.

Privacy Radar takes the same outside view to corporate customers. Available through Microsoft Marketplace since July 2026, it examines public digital behavior, estimates exposure and offers remediation steps. A company can investigate a mismatch between its privacy promises and what its live website actually does.

September’s Darrow Legal Alignment Benchmark gives that mismatch a concrete example. The opt-out defect occurred despite expert instructions. The study also found that clearer technical specifications improved compliance on its high-litigation parameters from 62% to 92.6%. These are results from Darrow’s test environment, rather than a universal ranking of coding agents.

The copyable lesson is specific: test the behavior behind the interface. A consent screen can look respectable while the network requests tell another story. Better instructions helped in the study, but they did not eliminate every defect. For insurers, that opens a question about measurable software risk; for compliance teams, it supplies something concrete to fix.

Growth has a human cost

Darrow raised a $35 million Series B led by Georgian in September 2023 to expand its team, legal domains and models. In August 2026, it announced Inc. 5000 recognition with 250% three-year revenue growth. A month earlier, CTech reported 60 layoffs from a workforce of roughly 180, including many legal analysts. Darrow said it had been profitable for three consecutive years.

Its published culture emphasizes people, candor and experimentation. The founders once kept an office sign mistakenly reading “You Get This.” It became a motto. Charming, certainly; the layoffs are a harder piece of the same company portrait.

Darrow’s promise depends on evidence being observable and interpretation surviving scrutiny. Private facts can remain hidden. A potential claim can fail on merits, procedure or economics. The useful product is therefore a better starting point for judgment. Whether the customer intends to bring a case or prevent one, the next move belongs to people.