Three former cybersecurity researchers walk into the legal industry. The punchline, apparently, is a network of AI-powered plaintiff law firms.
Erinys, a San Francisco outfit that emerged from Y Combinator's Summer 2026 cohort, announced August 11 that it had built what it claims is the first AI-native plaintiff-side litigation law firm network. The three-person team isn't selling software to lawyers. They're offering to run entire law firms for them.
The pitch is sweeping: Erinys will supply the technology, the support staff, the client pipeline, the billing systems, even the working capital. Attorneys who join get what the company calls a "turnkey" practice. Show up, supervise the AI, make the legal calls. Everything else gets handled.
"AI should produce the first pass," wrote Matthew Pallan, co-founder and CTO, in the launch announcement. "Lawyers should direct, review, decide, and advocate."
It's an audacious model in a profession notoriously resistant to disruption, though perhaps less audacious than it would have seemed a few years ago. Erinys builds its own technology rather than licensing existing tools, deploying the systems inside the practices it operates and iterating as problems surface. The AI handles initial legal research, analysis, drafting. Lawyers retain final authority over strategy and decisions, at least in theory.
What Erinys won't say publicly: how many firms have joined, what the revenue split looks like, whether lawyers take equity, which states it operates in. The company's Y Combinator profile listed the team size at three as of late August, though LinkedIn lists 2–10 employees.
An unlikely legal pedigree
None of the founders practiced law, which may or may not be the point. CEO Sasha Zyuzin comes from offensive security. He presented vulnerability research at Black Hat USA last year and spoke at the SANS OSINT Summit, after leading an 1,800-member cybersecurity club at the University of Maryland. Co-founder Jason Guo's background spans industrial control systems to kernel exploitation; he's finishing a computer engineering degree at Maryland. Pallan engineered trading systems for high-frequency firms that operate in sub-290-nanosecond timeframes, then aced the LSAT with a perfect 180 while studying computer science and real estate.
That technical firepower may explain a curious mention in a community writeup from a Washington, D.C., AI meetup earlier this year. The writeup described something called "erinys-investigate," a system that autonomously queries 22 U.S. government databases in response to plain-English fraud tips, complete with a chain-of-custody audit trail. Erinys hasn't publicly detailed this system, and the description comes from a community event writeup rather than official product documentation.
The competitive thicket
Legal AI has moved well past the experimental phase, and the business models proliferating around it vary wildly. EvenUp, which generates demand packages for personal injury cases, closed a $150 million Series E in October at a valuation north of $2 billion. UK-based Lawhive operates as a regulated legal services provider rather than a tool vendor. It raised $60 million in February and reported around $35 million in annual recurring revenue for last year, according to Fortune.
Y Combinator alone has seeded multiple legal AI experiments. General Legal, from the Winter 2026 batch, also calls itself an AI-native law firm. Moritz, an earlier entrant, pulled in $9 million last May to build an AI-driven corporate practice, Global Legal Post reported. Then there are the tool providers targeting plaintiff work specifically. Irys and Supio sell into existing mass tort shops rather than running firms themselves.
Erinys occupies a strange middle ground. Not quite a software vendor, not quite a traditional law firm, not quite a managed services provider. The model assumes lawyers want to practice but not operate, that AI can compress the economics of plaintiff litigation enough to make the arrangement viable, and that regulators in various jurisdictions will permit the structure. Those are substantial assumptions.
The company is recruiting. "We're looking for exceptional lawyers," the launch post read, directing inquiries to [email protected]. What exceptional means in this context remains undefined. So does the financial arrangement those exceptional lawyers might expect, or the volume of cases Erinys can funnel their way, or how the firm navigates unauthorized practice of law statutes in states with strict rules about non-lawyer ownership.
For now, Erinys is a bet that legal services can be reengineered from first principles by people who've never billed an hour. The industry has seen that bet before. Sometimes it pays off.
