Legal tech writing loves a good demo: instant redlines, smarter clause suggestions, fewer 2 a.m. deal scrambles, cheaper lawyer fees. But “AI-native” legal service providers revive an older, nerdier question: corporate structure. When a company calls itself an AI-powered law firm, is it one entity, two entities, or something in between? And if the answer is “two,” it’s hard not to think back to Atrium.
Crosby.AI (often branded simply as Crosby), fresh off their $60M Series B financing, is a prominent new entrant. It calls itself an “AI-powered law firm” for high-volume commercial agreements (NDAs, MSAs, DPAs) with fixed pricing and fast turnaround. The reported model is straightforward: AI handles much of the first pass, and a lawyer owns the review and final sign-off, backed by the usual table-stakes (a real firm, malpractice coverage, and pricing that’s more product than timesheet).
What I’m most curious about is the setup that makes that possible, especially in a U.S. market that generally limits outside ownership of law firms.
The “two-entity” question: where does the tech company end and the law firm begin?
Once you mix “venture-style” software building (models, engineers, product) with a regulated profession, one question pops up fast: if there’s outside capital involved, where does it sit? In other words, what part is the law firm, what part is the tech company, and how do they relate?
That’s not a “gotcha.” It’s basic diligence: who’s providing the legal service, who owns the tech, and how the two are connected.
Atrium is the closest comparison (and a useful cautionary reference)
Atrium is the closest recent precedent because it made the dual-entity model mainstream in startup circles. Launched in 2017, Atrium set up (1) a venture-funded tech company building software and (2) a separate law firm delivering legal services using that software.
The split was largely about compliance: in most states, nonlawyers can’t own a law firm or share in legal fees. So the investment typically goes into the tech company (equity-friendly), while the law firm stays lawyer-owned. Reporting on Atrium described the tech side funding platform development and helping finance the law firm’s overhead by making loans to the law firm through intercompany agreements.
Atrium later shut down its startup operation in 2020. The takeaway isn’t “this can’t work.” It’s that the model has built-in tensions:
Venture timelines vs. legal services reality: trust, relationships, and risk management don’t always scale like software.
Unit economics: even with automation, lawyers still need to supervise high-stakes calls.
Regulatory lines: the structure has to preserve lawyers’ independence while the tech company pursues growth.
So is Crosby “like Atrium”? Potentially in plumbing, but not necessarily in scope. Atrium aimed broad at the startup legal stack. Crosby’s positioning is tighter: high-volume contracting, fixed fees, fast turns, with AI doing first pass and a lawyer accountable for the final work. That narrower wedge may matter more than the label. However, I suspect that as Crosby grows, so will their service offering.
Why the two-entity model exists: nonlawyer ownership
Why the corporate structure gymnastics? Because in most U.S. states, nonlawyers generally can’t own law firms, and law firms generally can’t share legal fees with nonlawyers. There are potential exceptions (D.C., and newer reforms in places like Arizona and Utah are at least starting down that path), but the baseline is still lawyer ownership and control. Thus, where are the VCs to put all their investment money?
It’s possible Crosby uses that familiar split: a lawyer-owned firm delivers the legal services, while a separate company owns or operates the technology. If so, it’s structurally similar to Atrium, even if Crosby’s wedge (and the underlying AI) is different. In many jurisdictions, that separation may be the cleanest or most defensible way to build fundable software alongside a lawyer-controlled practice. In other words, since VCs can’t invest directly into law firms, they’ll invest in the adjacent tech company.
The more useful “Crosby vs. Atrium” angle is practical. Atrium paired big ambition with a complicated build. Crosby, at least as positioned today, looks like a tighter wedge: standardize a narrow contracting workflow and sell it at a predictable price. Either way, the corporate structure is doing real work, separating the entity that can take outside investment from the entity that can deliver legal advice.
Crosby is also a reminder that “AI law firm” is often a corporate design pattern: a regulated professional practice next to a fundable technology business. Atrium popularized that pattern in startup circles almost a decade ago. Crosby may be the next iteration and less “law firm of the future” theater, more a focused attempt to make the two-entity model more operational.
Why I’m paying attention
Personal note: I own venturelaw.ai and startuplawyer.ai. That’s part of why I’m interested in how Crosby’s structure is put together and, more broadly, in understanding how these “AI-native” firms actually operate in practice (from fully automated workflows to lawyer-plus-ops teams behind the scenes).
I’m cheering for the outcome…and still curious about the plumbing.
Ryan RobertsStartup Lawyer
Ryan Roberts is a startup lawyer at Roberts Zimmerman PLLC with more than two decades of experience advising startups and venture capital investors. He is the author of “Acceleration” and StartupLawyer.com.
