Crosby: The Inner Workings of a Neofirm — New Ontologies
An immersive look inside Crosby, an AI-native law firm achieving 10x productivity gains while reimagining professional services: fixed fees replace billable hours, lawyers run evals alongside engineers, and the entire business model gets rebuilt from scratch.
Read Original Summary used for search
TLDR
• Crosby lawyers redline NDAs in 4:53 vs 58 minutes traditionally, but the real innovation is structural: fixed fees vs billable hours, corporate R&D reinvestment vs partnership profit distribution, lawyers commanding AI agents rather than being commoditized
• Technical edge comes from custom eval systems, living playbooks that encode client preferences over time, and infrastructure for long-running stateful agents—moving toward game theory models of negotiation
• Culture is the moat: engineers shadow lawyers for weeks to build empathy, lawyers run evals and feel like they're "commanding an army of agents," everyone eats lunch together despite mixing elite law firm partners with 22-year-old engineers
• The "neofirm" model could be a blueprint for all professional services transformation—success requires making experts feel empowered by AI rather than threatened, preserving human judgment while automating tedium
In Detail
Crosby is an AI-native law firm that's fundamentally restructuring how legal services work. Founded by Ryan Daniels (Stanford-educated attorney) and John Sarihan (early Ramp engineer), they've raised $85M and serve customers like Ramp, Cognition, and Cursor. The productivity gains are dramatic—lawyers redline NDAs in under 5 minutes versus 58 minutes traditionally—but the deeper story is about reimagining the entire business model and culture of professional services.
The technical challenges are fascinating because law is a non-verifiable domain where two smart lawyers can legitimately disagree. Crosby is building custom eval systems that automatically funnel failure cases, living playbooks that encode evolving client preferences (remembering instructions from Slack messages, creating game theory models of what clients will compromise on), and infrastructure for long-running stateful agents that can't just crash and retry. The long-term vision is using reinforcement learning to make contract review more verifiable, eventually becoming "contract infrastructure for the internet."
But the real innovation is organizational. Traditional law firms are partnerships that maximize partner profits and treat junior lawyers as "operating expenses to be optimized." Crosby is a corporation that reinvests profits into R&D, charges fixed fees instead of billable hours, and creates what they call a "lawyer utopia"—where lawyers only do expert work while AI handles tedium. The culture is deliberately designed to make lawyers feel they're commanding AI agents rather than being replaced: engineers shadow lawyers for weeks to build empathy, lawyers run evals and tune models, everyone sits together and eats lunch as a team. As one lawyer put it: "That feeling that I'm commanding an army of agents, if you can create it, feels amazing. That feels empowering."
The piece argues this "neofirm" model—equal parts elite domain experts and top engineers, owning full service delivery, obsessed with both technology and human relationships—is a blueprint for how all professional services will transform. The winners won't just have the best AI, but will solve the harder problem of making professionals feel they have agency in an automated world.