Ryan Daniels

The through-line

Ryan Daniels' recurring argument is that law is one of the most consequential institutions in society and also one of the least changed by technology, and that closing that gap requires becoming a law firm rather than selling software to one. He grew up with the conviction, repeated constantly at home, that "the legal system is really important" [1], and separately with a fascination for what computers do to cost structures: "you can take this thing that required a lot of humans and had a huge marginal cost to do before variable cost and now you can just do it at scale" [1]. For years those two interests sat awkwardly together. What joined them was the observation that lawyers are "an unbelievably important part of society, especially US society. And it's overwhelmingly manual the work we do" [1].

The shift over time is from theory to urgency. At a Stanford think tank in what he places around the GPT-1 era, the idea that software would eventually change how lawyers work was "very theoretical cuz nothing was really that useful" [1]. The models made it concrete, and the conviction has since hardened into an operating thesis he now describes as automating a trillion-dollar legal industry while keeping lawyers central to the delivery [3][4].

On the moment the technology became real

The founding trigger was using ChatGPT. "It was almost emotional" [1], and while GPT-3 "wasn't amazing at legal work," 3.5 was the point where "this is really going to change the way legal work happens" [1]. The force of the moment came from contrast with his own experience as a corporate associate at Cooley: he realised "I was probably working the same way my parents were more or less, maybe just with email from so many years before" [1]. He is candid that the insight was not unique, and suspects many people had the same aha moment, but that for him "it was hard to put it back in a box," and afterwards "I've never felt such a sense of I can't do anything else" [1].

He also frames it as the arrival of something long anticipated rather than a surprise: "the thing that we've been talking about in law school is finally going to happen" [1].

On why selling tools to lawyers wasn't enough

Having left the startup where he was GC, he spent roughly nine months talking to more than a hundred lawyers through his law school network, asking how they were using the new models and looking for "pockets of law" worth focusing on [1]. What emerged was a claim about buyer psychology rather than product capability: "when people have some sort of legal issue, their instinct is to go to a trusted advisor. That's what lawyers really do. Not necessarily to like an AI system" [1]. He expects that to change eventually, but the founding decision was "let's just meet people where they are" [1].

Investors were unconvinced. Many told him it was "a cute idea, but like it wouldn't really work" and pushed him to be "a bit more software oriented" [1]. The market evidence that settled it came from digging into alternative legal services, the largely offshore alternative to law firms, which he researched partly by spending time in India: it is a bigger market than legal software and growing four times as fast [1]. There was also an operational reason to own the firm. To move real work onto agents, "we had to have lawyers working with us, for us, under one roof," with incentives strong enough to change behaviour week over week [1].

On killing the billable hour

Daniels treats pricing as the mechanism that makes AI adoption inside a firm possible at all. The billable hour defeats it arithmetically: "if you do 10 tasks in an hour, 100 tasks in an hour, you're still billing an hour" [1]. Crosby bills by the piece of work instead, and he is explicit that this was chosen knowing "we'd actually run the risk of being margin negative on some days of work," because the incentive it creates is "really profound for everybody in the company to just rally around this goal of we need to find a more productive way to work" [1]. The demand side validated it quickly: he started with companies carrying legal budgets in the hundreds of millions, then found that investors' offices and founders wanted the document-priced model immediately, which he describes as "a very organic way of finding your place to start" [1].

On contracts as the moment that matters

The wedge is narrow by design. "Every startup, every company in order to generate revenue needs to negotiate contracts. That's the moment that matters most" [1], and the value proposition is speed at that moment, with Crosby reporting cycle time improvements in the range of 70 to 80 percent in some cases [1]. He points to clients where deal velocity and contract quality are mission critical, and to volume as proof: over a billion dollars of client contract revenue closed in the 270 days since the seed announcement, against roughly 20 or 30 million at the point of coming out of stealth [1].

On research and where the firm goes next

The next phase is framed as a research problem rather than a services scaling problem. The ambition is to model the interaction itself: "we're kind of just trying to basically predict the way two parties negotiate an agreement" [1]. The $60 million Series B, which brought in Lux for the first time as a deep tech partner with a New York team, is explicitly meant to fund that, with leaning more heavily into research described as the major focus for the coming year [1].

On outsider perspective and the pull toward building

Two formative threads sit behind the company. One is a family relationship with law that is unusually literal: seven or eight lawyers among the cousins, a law professor father, a human rights lawyer mother, and a grandfather who fled Eastern Europe before the Holocaust and entered Canada on a judge's ruling, which became family lore about the legal system as a route to agency in society [1]. Daniels notes the system is often misunderstood because "you only see lawyers sometimes when you don't want to see them" while it quietly structures ordinary life [1]. He was nonetheless ambivalent about practising, because he loved tech too much [1].

The other is a habit of pursuing contrarian ideas and rallying people around them, which he traces to childhood and to being the family IT guy in "the cult of Apple" [1]. He half-jokes that growing up in Canada, where everything is nearly but not quite the same as the US, means "everything looks just slightly askance and I think over time it just makes you question things just a touch" [1]. That instinct showed up early in his career: joining HireScore at about five people intending to do a summer internship and never leaving, working on AI for hiring "back when it was ML and data science," and ending up writing long policy memos to the EEOC and federal regulators arguing how such systems should be regulated and how liability should work [1]. That work led him into AI policy research and, indirectly, to the thesis he now builds on.

Takeaways

• The decisive product insight is behavioural, not technical: people with a legal problem go to a trusted advisor, so Crosby became a law firm instead of selling software to firms [1].
• Removing the billable hour is what makes agent adoption possible internally, because hourly billing pays the same whether a lawyer does 10 tasks or 100 [1].
• Daniels accepted the risk of being margin negative on some days in exchange for company-wide incentive alignment around productivity [1].
• The market argument for legal services over legal software: alternative legal services is a larger market than legal software and growing roughly four times faster [1].
• Traction cited: over $1 billion in client contract revenue closed in 270 days, up from about $20 to $30 million at the seed announcement, alongside a $60 million Series B with Lux joining [1].
• The research agenda is to predict how two parties negotiate an agreement, and that is the stated focus for the year ahead [1].
• The founding realisation came from seeing that as a corporate associate he was working essentially the way his parents did, plus email [1].