The launch speed of a startup depends on how quickly they can develop technology, but also on something seemingly more mundane: how quickly they can sign contracts.

Contracts are the lifeblood of a business, and one of its most persistent sources of pain. They’re still priced by billable hour, still largely done by hand, and still costing companies weeks of back-and-forth on deals that should close in days. And delays of even a day can add up, costing billions a year. We saw this up close with companies in our portfolio spending weeks on contracts stuck in review, watching deals stall while lawyers racked up charges.  

That’s the problem Ryan Daniels set out to solve when he co-founded Crosby, a registered AI-native law firm that reviews and redlines contracts in under an hour, pairing large language models (LLMs) with in-house lawyers who take on direct liability for their work. In the past year, they’ve grown from 0 to over 100 customers – including Cursor, Ramp, Tishman Speyer, and Starwood – processing tens of thousands of contracts and billions in transaction value. Today, we’re thrilled to announce that Lux is leading Crosby’s $60 million Series B.

I first met Ryan in late 2023 in New York. He co-founded Crosby with John Sarihan, formerly a top engineer at Lux portfolio company Ramp who had watched the company scale and felt legal drag firsthand. In fact, there are key parallels between Ramp and Crosby, with both championing their customer. In Ramp’s case, it’s the CFO tirelessly closing the books as company financial complexity scales while in Crosby’s, it’s the in-house general counsel closing contracts as legal complexity skyrockets.

They’re both Penn alums — John was part of the Penn M&T program that Lux has gotten to know well over the years. What struck us early was a story Ryan told about a trip to India, where he’d spent time observing the legal outsourcing industry. He saw an entire back-office ecosystem designed to serve American law firms where high billable hours precluded working on the overflow of commercial contracts instead of nuanced strategic input. Businesses had been routing around the white-shoe rates for years. Ryan saw that as proof of demand, and AI as the means that could finally close the quality gap.

What convinced us was when Crosby publicly launched last year and our portfolio companies were clamoring to use the product. By the time they signed their third Lux portfolio company and were recruiting some of the best talent in NYC, we’d seen enough.

Legal AI is crowded, but the distinction that matters most is often overlooked. Most companies in this space sell productivity tools to law firms, analogous to in-line code completion today — speeding up what lawyers already do while leaving the work to humans. Crosby takes a fundamentally different approach: what Ryan calls a “neofirm”. It’s a services company built from the ground up to integrate the workflows of professionals and AI agents. Crosby isn’t an AI assistant for lawyers but an AI-accelerated law firm in its own right, replacing traditional legal services — one that assumes direct liability for the finished contracts it delivers.

Here’s how it works. You send a contract to Crosby over Slack or email. It could be a Non-Disclosure Agreement (NDA), Master Service Agreement (MSA), Data Processing Agreement (DPA), or something else. Its AI suggests changes or examines changes proposed by your counterparty. While some are routine, it flags trickier ones and routes them to Crosby’s lawyers, along with explanations. Those lawyers weigh in, and Crosby returns your redlined contract, typically in under an hour. 

Take Cursor. Before Crosby, a sales rep waiting on Fortune 500 redlines could wait days for their law firm to respond, directly risking deals and stalling time to revenue. With Crosby, the same contract comes back in hours. Crosby has processed over 2,000 contracts for Cursor alone, cutting review time by 50%. That’s a huge advantage over using a legacy law firm. Just as radical as the speed is the billing. Crosby bills based on outcomes instead of billable hours, so their incentives are aligned with customers.

Crosby’s durable advantage is how the tech compounds. At Lux, we have a long history of partnering with some of the best technical builders across AI from Hugging Face to Together AI, Cognition and Modal. LLMs are tuned using a training process called reinforcement learning (RL), where human feedback or verified signals reward the model for better outputs. For math and code, RL can largely be automated because you can objectively check whether an answer is right. Legal decisions aren’t verifiable that way; whether a redline is "good" depends on context, preferences, and judgment. 

To solve that problem at scale, a product needs to do two things. First, it needs to decompose the problem into smaller verifiable tasks and second, it needs to build a golden dataset of what “good” looks like, which requires a lot of data. That’s what makes Crosby’s approach so technically interesting. Their lawyers don’t just oversee the AI; they improve it as they go. Crosby has built a unique reinforcement loop that gathers real-time feedback from in-house lawyers and customers, so the model gets better with every interaction — adapting to a company's specific contract types, preferences, and negotiation patterns.

The advantage of a neofirm is that as AI assists professionals, professionals return the favor — a virtuous circle that builds an increasingly hard-to-replicate data moat. Longer term, Ryan and John see Crosby as sitting at the center of predictive negotiation. When two parties can train AI models on their preferences and legal agents negotiate with each other automatically, they can reach agreements faster than any human-to-human negotiation could. Both parties get what they want, and the lawyers don’t rack up their fees in the process.

The billable hour has defined the industry for decades. Crosby intends to replace it. That takes elite lawyers who want to reinvent their field working side by side with AI researchers productizing real-world feedback loops. If this mission resonates with you, they’re hiring.