Crosby raises $60M Series B and crosses $1B in client contracts using AI agents to run a law firm end-to-end
Mar 31, 2026 with Ryan Daniels
Key Points
- Crosby raises $60M Series B and has closed over $1B in client contracts since launching, with funding directed toward building specialized AI models as the company scales.
- The startup replaces routine associate work with AI agents rather than eliminating lawyer roles, accelerating junior lawyers into relationship and deal work that requires human judgment.
- Crosby plans to shift from relying on frontier models from OpenAI, Anthropic, and Google to proprietary fine-tuning within roughly 18 months as it accumulates competitive data.
Summary
Crosby, a startup using AI agents to run law firm operations end-to-end, has raised a $60M Series B and crossed $1B in client contracts. The funding will go toward building a research team to develop more specialized models as the company scales.
The commercial logic behind the bet starts with a striking profit figure: the top 100 law firms collectively earned just under $70B in partner profits in 2025 alone — more than Google spent on all its R&D that year. Crosby's founder argues that redirecting even a fraction of those profits into better tooling and client experiences would expand the legal market rather than compress it.
Where AI fits in the org chart
The near-term model isn't full attorney replacement. The founder, a Stanford Law graduate, sees two roles surviving and growing: client-facing work that requires interpersonal judgment, and the ability to communicate legal reasoning clearly to engineers building the tools. Both require actual lawyers — which means firms still need to hire, but differently.
For associates specifically, the argument is that AI accelerates the career path rather than eliminates it. Rather than being buried in routine paperwork, junior lawyers can move faster into the relationship and deal work that AI can't yet navigate.
Model strategy
Crosby currently runs on frontier models from OpenAI, Anthropic, and Google, relying on well-structured context and agent workflows rather than proprietary fine-tuning. The founder draws a parallel to code-generation companies that got significant early lift from base models before eventually needing to fine-tune as they accumulated data and faced competitive pressure. Crosby expects to go that direction within roughly 18 months, which is part of what's driving the pace of this fundraise.