Erinys starts with a lawyer who wants to start and own a firm.

Erinys provides the operating infrastructure around the practice: taking on clients, AI-assisted research and drafting, the back office and marketing. The lawyer owns the practice and makes every legal decision. The operating infrastructure around them comes from Erinys.

Read that as a product and it is a decent piece of legal software. Read it as a change in how the market is arranged and it gets much more interesting, because it says something uncomfortable about why law firms exist at all.

What they do

Erinys describes itself as an accelerator for AI-native law firms, building a network by giving lawyers the infrastructure to start firms of their own. It supplies the technology, the client sign-up, the back office and the business development a practice needs. AI does the first drafts, the research and the analysis, while the lawyer keeps control of judgement, advocacy and every legal decision.

Erinys is deeply committed to making legal services more accessible, including by working with nonprofit organizations.

The customer is not the client with a legal problem. It is the lawyer who wants a practice.

It is based in the San Francisco Bay Area. Sasha Zyuzin, the chief executive, is a cybersecurity researcher who has presented at major industry conferences and researched vulnerability discovery and malware analysis at the University of Maryland, where he also taught open-source intelligence. Matthew Pallan, the chief technology officer, previously worked on hardware for high-frequency trading systems; before that, he built and sold cybersecurity products. He also scored a perfect 180 on the LSAT. Jason Guo, the third founder, works on industrial control systems and kernel exploitation.

Who else is in this field

Legal software is crowded in two directions at once. There are the tools sold to existing firms: Clio for practice management, Ironclad and Spellbook for contracts, Everlaw and Relativity for discovery, LexisNexis and Casetext for research, Harvey selling AI into large firms, and EvenUp working personal injury specifically. And there are the ones that go past the lawyer straight to the client: LegalZoom, Rocket Lawyer and DoNotPay.

The warning sits between the two. Atrium, the tech-driven law firm founded by Justin Kan, raised a lot of money on the idea that a firm rebuilt around software would outrun traditional practices, and closed in 2020. Erinys is going after a similar position with one difference that did not exist then: the drafting and research it automates is work that software genuinely could not do six years ago.

The same breed as its batchmate, pointed differently

Erinys belongs to a type of company that barely existed a year ago and turned up several times in this one batch: companies that are not selling software to a profession, but rebuilding the basic unit of that profession around AI.

The logic is the same one an accounting firm in the same batch is running. The AI is not being asked to beat a trained professional. It is being asked to match them on the work around the edges, which is a much easier target, and then to remove the staff that work used to need.

What that produces is a whole working machine rather than a tool. Sign-up that answers at any hour. Correspondence that never falls behind. Research and first drafts. Billing, scheduling, chasing, and a steady flow of new clients. None of it is legal expertise. All of it is the dull machinery that takes years to build, costs a fortune to staff, and is the real reason running a practice is hard.

Sold as one working package, that machinery is worth more than any single piece of it, and it is the part one professional cannot put together alone.

The opportunity underneath: lawyers who want to go out on their own

Lawyers have always wanted to start firms; the hard part was everything around actually running one. Starting a practice meant building intake, hiring staff, setting up billing, insurance, technology, marketing and a way of finding clients, all before the firm had any real scale.

For most lawyers, joining an existing firm was the practical way to get access to that machinery.

If Erinys can provide that infrastructure around an individual lawyer, starting a firm becomes realistic for far more of them.

Erinys is aimed straight at that group: lawyers who want to build something of their own.

Why this is disruptive rather than merely useful

Consider what happens if it works even moderately well.

Traditional firms often make money through leverage: clients pay for legal work performed across teams of lawyers at different levels of seniority. Give more lawyers a credible path to ownership, and some of the economics that hold that model together begin to change.

At the same time, the economics of which matters are worth taking begin to change. In contingency-fee practices, the issue is not pushing prices down: clients generally do not pay legal fees upfront. The constraint is whether a matter is economical for a firm to pursue. If AI reduces the cost of intake, research, drafting and administration, smaller or previously uneconomic matters can become viable for a new firm.

It is genuinely disruptive, and the interesting thing is that it does not require established firms to adopt the model. Individual lawyers can choose to start AI-native practices of their own.

Why it is bigger than it sounds

The short description is an accelerator for AI-native law firms. What matters is what that makes possible.

  • More lawyers can build firms of their own. Erinys gives an individual lawyer access to infrastructure that previously took time, money and a larger organisation to build.
  • More matters can make economic sense. Lower operating costs can make matters viable that a traditional firm might not have been able to justify taking, particularly in contingency-fee practices.
  • A law firm can become much smaller. Technology, intake, research, drafting and back-office infrastructure can increasingly sit around a single lawyer rather than requiring a large organisation.

What to watch

The number that settles this is how many practices actually start on it, and how many are still going a year later. A network of AI-native law firms is either a real network or a slide in a deck, and the difference will show quickly.

The second thing to watch is who the lawyers are. If they are experienced people leaving established firms, this is a better way to run a practice. If they are people who would otherwise have taken a salaried junior job, then something deeper is happening to the profession, and the same pattern is available in every licensed trade where the qualification belongs to a person and the overheads are what stop them going out alone.