While everyone watches Harvey and Legora fight for the generic legal AI crown, the real money is going vertical.

The Summary

  • Patlytics raised $40M Series B led by SignalFire to automate the full patent lifecycle, from filing to litigation
  • The same AI model that predicted Harvey's unicorn status in 2023 (now worth $11B) flagged Patlytics in March 2026
  • Thesis: Horizontal legal AI platforms hit a ceiling, vertical specialists win the complex work

The Signal

Patent law is where the agent thesis either proves itself or dies. Not because it's glamorous, but because it's brutally specific. Patlytics handles invention disclosures, portfolio management, and IP litigation, tasks that require understanding claim language, prior art searches across decades of filings, and the kind of procedural nuance that breaks general-purpose models. CEO Paul Lee's bet is that Harvey, despite its $11 billion valuation, can't easily replicate what Patlytics does. He's probably right.

The funding round tells the story better than any pitch deck. SignalFire led, but look at the participants: Jeff Hammes, former chairman of Kirkland & Ellis. Antiportfolio Ventures, started by a former Kirkland lawyer. Relativity, the legal data intelligence giant. These aren't AI tourists. They're IP insiders betting that the future of legal work is unbundling the law firm, one practice area at a time.

Lee came out of Peter Thiel's Mithril Capital, then Tribe, before founding Patlytics. His investor lens spotted what others missed: IP disputes mostly settle quietly, creating a "massive secretive market" invisible to most tech founders. That's the edge. Not better models, but better domain knowledge about where the work actually happens. Most patent fights never see a courtroom. They're resolved in negotiation, portfolio analysis, claim construction. That's where automation delivers ROI.

The TRAC prediction model adds weight. In 2023, it flagged Harvey when the company was less than a year old. Harvey hit $11 billion. Now it's pointing at Patlytics. Whether TRAC's model is signal or noise doesn't matter as much as the pattern: vertical AI beats horizontal AI when the work requires deep expertise, not just speed.

The Implication

Watch for more legal verticals to get funded. Tax law, M&A due diligence, regulatory compliance, each complex enough to resist commodification by Harvey or Legora. The agent economy won't be won by the biggest platforms. It'll be won by the ones that know which 10,000 hours of human expertise to compress into software. If you're building agents, go narrow and deep. The IP lawyers are showing the way.

Source: Business Insider Tech