The Crosby Story – With Co-founder, Ryan Daniels

Crosby was one of the original AI-first / NewMod law firms to come to market and it is growing fast. Here, Artificial Lawyer talks to co-founder Ryan Daniels about the origin story of the pioneering legal business, the personalization of legal AI outputs, attracting top tier lawyers, the use of agents and how they will evolve, plus its long-term strategy and more!

To watch / listen to the interview, please press Play, or you can go direct to the AL TV Channel.

We really had a great conversation and what struck AL is that Daniels does not see Crosby as providing commoditized legal work – far from it. They employ experienced lawyers, and the use of AI, in particular in reference to a client’s particular playbooks and style, is all about providing a customised and quite sophisticated end product. I.e. this is not a cookie cutter approach, it’s just designed to be way more efficient than other, more traditional, methods.

Their approach allows for each and every lawyer on their team, and each and every client, to come together to have contracts handled in a way you’d expect to see at a major law firm. It’s just that they will be far faster, and far less expensive, and operate on a fixed fee.

And on the fixed fee point, Daniels is open about the fact they sometimes make a loss on a piece of work, something most Big Law firms would never accept. But, for Crosby, it’s all part of building what to them is truly the ‘law firm of the future’, or what AL calls a ‘NewMod’ or new model law firm, where the lawyers and the AI are both put to maximum use, and neither prevent the other from doing what they need to do because of the time-based business model of traditional firms.

We also explore how they got started, how the model works, how NewMods will change the market, and much more. Enjoy.

There is also an AI-generated transcript below.

More about Crosby here.

—

AI Transcript of AL TV Interview

RT – Hey everybody, Richard Tromans here again, Artificial Lawyer TV, doing a special interview. This time we’re talking to Crosby and in particular, we’re talking to co-founder Ryan Daniels. Hey Ryan.

Ryan Daniels (00:16.991)
Richard, it’s so good to be here.

RT (00:18.88)
And it’s the first time we’ve actually spoken, even though I’ve actually written about Crosby several times now. So it’s great to actually meet and record this video. So first of all, let’s start off with the origin story. How did Crosby get going?

Ryan Daniels (00:34.259)
It’s so good to finally talk to Richard. We across the company are very big fans, read a lot of what you put out. So, you know, Crosby, the idea starts in some ways a few years ago, and in some ways like many, many years ago. my parents are legal academics. I’ve always been very interested and drawn to just the sort of hidden pervasiveness of the legal system in the US and how it, you know, it it just affects so many more things in our life than we appreciate. I always loved tech. But

You know, when I was at law school, I happened to be at Stanford, right? When the transformer models came out and, you know, kind of language models were were taking off. I’d worked in AI before law school. And so I think I just started to see that these things at some point would be on a perfect collision course. And I hadn’t fully grasped this was twenty seventeen, twenty eighteen, how quickly language models would progress. But in twenty twenty-two, twenty-three, when you know GPT-3 came out, it was obvious this was gonna be a new frontier for legal and the most interesting paradigm.

The origin story for Crosby was, you know, I think at a certain point I just became completely transfixed by, you know, I think even between GPC three and 3.5, how quickly things were improving. And I knew that law would be the service area that probably second to code generation would be most impacted by large language models. And talking to as many lawyers as I could, I think I spoke to about a hundred in in in a couple months. I I’d left my last job. I was just focused on starting something. I kept hearing

RT (02:01.51)
You’d worked as a lawyer in between this period.

Ryan Daniels (02:04.576)
Yeah, that’s right. So I I started my career cooly at you a big law firm and then I was a G C at a startup and we were growing quite fast. And so but the whole time I was kind of like always fascinated by how AI would be helpful. And I just kept hearing from lawyers that, you know, AI was useful, but they were always checking the outputs, that the judgment of, you know, their lawyers mattered the most. And so even if AI could be somewhat helpful, they always had to review everything unless a lawyer looked at it. and

You know, at one point along the journey, I got tipped off by a couple of CLOs to offshore legal services that, you know, they were spending moving a lot of their spend from big law firms to. And I spent some time in India digging into it. And so I was just seeing this like, you know, the way the ecosystem was looking was AI was getting pretty good. people were desperately looking for alternatives to big law firms. and the judgment of lawyers looking at work was really important. So the more we were kind of looking at this, we thought, well, it’s the hard thing to do, but the obvious thing to do is start your own law firm.

align the business model where there’s a real incentive to offload work to agents. And as AI gets better and better, which I took as just axiomatic as it would, the law firm sort of business model gets gets better and better, right? Because you get more lift from agents. And so my co-founder John and I met and we’re working on this idea for several months. And that’s what we ended up doing.

RT (03:18.957)
Fantastic, fantastic. And did you sort of, did you kind of like stumble upon this idea of a sort of MSO model and, you know, as you know, and then you realize, yes, as we become more efficient, you know, we’re like Henry Ford making his factory more efficient, you’ve got the mechanized bit and you’ve got the human engineers as well to finish the product. I mean, did you, did you just kind of like, was this like a one step leading to the next?

Or did you get like a kind of like gestalt moment and you just went ding, I see it.

Ryan Daniels (03:51.743)
You know, it was like these the the sort of pre-company formation like of the idea it’s so nonlinear. Like you have this idea and you’re like, That’s cool, but that’ll never work, and you just you drop it, then you come back to it. So it was very circular or sort of just nonlinear. And you know, the sort of like the axiomatic the the the great idea for us was like, well start our own law firm, but we kept being like, nah, like okay, we’re gonna have to hire lawyers, incorporate a law firm, like deal with the bar, like

That’s not easy to scale. And it’s, you it’s just there’s just too much hard stuff. And you we were talking to VCs and they’d be that’s insane. You know, like we’re gonna back that. But like the more we’d circle back to it, the more we kept seeing like it just solves a lot of our, it’s really hard to do. You know, and like, you know, you look at the big hundred law firms in you know the US, they’re most of them are over 70 years old. Like we were just kind of like, what are we thinking? But the more we’d circle around it, we’re like, it just solves so many problems, right? It solves the incentive problems, it solves the business problems, it makes it easier to package for the clients.

And so I think there was a moment where John and I looked at each other and said, I think we just gotta do it and and you know, that’s the swing we’re gonna take.

RT (04:52.429)
Yeah, yeah, yeah. And you are using an MSO model, is that correct? Actually, for some people, mean, a lot of people know this who are regular readers, but for some people who may not know what that is, could you just maybe explain it a little?

Ryan Daniels (04:56.274)
Yeah, that’s right. Yeah.

Ryan Daniels (05:04.67)
Yeah, totally. Well, I’ll say like in the US, there’s a few models that were floating around at the time. There was the ABS model, which is this alternative business structure, essentially allowed non-lawyers to own equity in a law firm entity, which is, you know, in every other state. or or had to have fee sharing with non-lawyers in in a in a law firm. The other popular model was to have basically two entities, one being a law firm, one being a corporation or non-law firm, and have basically a partnership or agreement between the two.

or service agreement between the two entities whereby the corporation sells and licenses technology and some other services, maybe right invoicing, marketing, sales to the law firm. And the law firm provides legal services to a bunch of clients, but has this arrangement where they’re buying a bunch of non-legal stuff from the corporation. And you know, if AI can do a lot of the work of lawyers, but let’s call it just paraleg, then you’re buying the law firm is buying all that from the law firm.

This model I would say was mostly pioneered by this company Atrium, you know, that was kind of the first AI law firm in many ways in twenty eighteen. And so, you know, we we we we we didn’t invent the wheel there. We just kind of dug deep into the business model. It made a lot of conceptual sense and so we just operationalized it.

RT (06:08.375)
Yeah.

RT (06:18.189)
interesting isn’t it’s interesting is like you know the legacy the legacy of know of of atrium lives on but perhaps not as they first intended but it’s sometimes it’s like that isn’t it it’s it’s you know sometimes the early pioneer doesn’t really make it but they set in motion a whole bunch of things that come later

Ryan Daniels (06:37.396)
You know, like so I ended up speaking to I think all of the founders of Atrium before we started Crosby and just kind of tracking them down. And almost all of them basically said it was the right idea. It w and but it was just that, you know, that like AI wasn’t there yet. And but you could see that the vision was really coherent. And you I I read some of like the vision documents they’d written and it was just it was totally right. And you could just see that if if it was written today, it would just be spot on.

RT (06:55.191)
Yeah, it wasn’t.

Ryan Daniels (07:07.218)
So it it is amazing. And you know, I I I’d I’d wish it I’d wish it I wish it had inspired more and more lawyers to kind of like see like what was possible. It was just a touch too early.

RT (07:17.229)
Yeah, yeah, yeah, yeah, yeah. We’ll have to do a separate thing about that, interesting, interesting. Well, let’s scroll forward to today. So some people will say, oh, Crosby, I think I heard about them. Don’t they just do NDA review? Right. Let’s bring everyone up to date. What are you doing now and what are you offering?

Ryan Daniels (07:35.371)
So we started the you know, we started the law firm by saying like what is the what are the like the pieces of legal work that companies will be most comfortable offloading to an AI law firm? And and then the simple question we asked was, you know, what is probably going to be most offensible from just like the models getting really good at it? And and kind of where can we play in that intersection? And it turns out that contract review, sort of writ large, just like you high-volume contract review, fits that mode for three reasons. One is

The liability profile is, you know, fairly remote, right? Like there are real problems from negotiating a contract poorly, NDAs aside, right? So now we’re doing MSAs, a lot of DPAs, and sort of partnership agreements, like slightly more complex agreements, we’re just moving up the complexity curve. but the way that AI reviews a contract is extremely judgment based. Like

You know, at at its core, it’s a negotiation between two people. And you’re you’re going back and forth and you’re trading like, you know, I’ll give you this if you give me that. And a very good lawyer understands that, you know, if I change the liability cap, I can give a lot of other things. And so it’s a very complex dynamic that, you know, AI alone is still not great at. And so we understood that clients should be really willing to hand it off to us. It’s easy to build a big data model around on a per client basis because there’s so many of them. And it’s still really hard for agents to do on their own.

And so that’s where we’ve only focused. As we’ve continued, again, we do these more and more complex agreements. so you know, especially around like the data privacy stuff, data buying, things are getting slightly more complex. we’ve also, there’s just a lot more you can do with that, right? Like once you have all that data, you can, you know, people are asking more questions about like all of their obligations to their clients and to their, to their, to their vendors. It’s all contained in those contracts. So there’s just more you can build on top of it.

RT (09:24.695)
So, okay, so, but it’d be fair to say that, suppose we could say this is transactional work in the sense that it’s contract as parties, but we’re not getting into like &A deals. We’re not getting into them.

Ryan Daniels (09:34.025)
Not yet. Like we’ll do sometimes, you know, we’ll get asked to help with diligence for an MA because we have all the contracts and so we can, you the stuff we can pull out. but yeah, we’ll we’ll we we haven’t gotten there yet. it’s it’s very much something that we’ve thought about, which is like, you know, be sort of financing, especially for our startup plans, venture financings, maybe some MA work, maybe some fun formation. but but you know, not yet. you know, focus on the one thing. Yeah. Yeah. And and and and I think to your point.

RT (09:58.882)
the onion layer by layer.

Ryan Daniels (10:03.508)
Again, the work is quite different and and I mostly did venture financing work when I was when I was at Cooley. The fundamental unit is a contract. And so are kind of think in you know, obviously there’s a lot of ancillary documents, but if you can get really, really good at building a system that can negotiate contracts, you really can see that it can branch out to more and more complex work.

RT (10:21.901)
Yeah, yeah, yeah. I mean, this is the thing. I mean, I’ve often wondered if the future of and I call them new mods and I know someone at the the recent Legal Innovators California conference called them neo firms, but then by the halfway through their talk, they became neo mods. I can go with neo mods because neo just means new, so I can live with that. And other people call them AI first law firms.

Ryan Daniels (10:40.552)
Ha ha ha ha.

RT (10:48.621)
I think that’s kind of, it’s a very cool way of describing them, but I think that eventually all law firms will become AI first. So it’s like, let’s call them something else, you know? But to that point, do, does, you you and there’s a whole bunch of others, we won’t talk about the others yet, because you know, we all know who they are. But are they gonna grow like you, like seeds, are they gonna start to grow outwards, or do you eventually combine?

Ryan Daniels (10:53.566)
Yes.

Ryan Daniels (10:58.026)
I agree.

RT (11:12.993)
with a bunch of the other ones and between you, you become a much larger law firm. I suppose it wouldn’t work if you all did the same thing, because you’re just going to be more of the same. It wouldn’t make any sense. if one was very private equity and another one was very much focused on business immigration law, you know, and you put all those together, you’d have a proper big law firm, but it would be a new mod at the same time.

Ryan Daniels (11:37.579)
So I think like, I mean y y this will be fun to discuss with you because I think you understand the you know the economics of law firm so deeply. Like the I think one of my one of our key insights quite early was, and we we you know, neo firms, neo mods, whatever, like the the the there’s always been there’s a there’s a famous paper from years ago that says why law firms collapse and it’s you know the idea is that when law f there hasn’t been

really serious consolidation in the law firm industry, right? And in the US, you know, there is no non-competes, so law firm so lawyers can go join another law firm and there’s no non-solicit, so they can take their clients with them because of the ethical obligations of the client, not the firm. And what that means is when a law firm gets big enough, there’s a real incentive for the nth plus partner to go join another firm and you know get a little bit more of the partnership and more upside. And so they can never get to a certain scale before they just get too top heavy and start tipping.

Kirklands may be defying that, but most law firms kind of have like a certain cap. Unlike in a canton where you the big four that just really consolidate. And the kind of just the theory that we’ve had is if the underlying system of, you know, the sort of agentic system that the firm is built on top of, which contains a lot of rich context about clients, you have agents that can do actually like a meaningful amount of the work. and the lawyers are really here to offer judgment, it becomes, you know, you actually could have a law firm that has

You know, that that that gets better with every additional lawyer and you know, like and sort of k you know, exponentially better. And so, you know, the incentive is to stay. And so you can actually have this one consolidated firm, and maybe it will be us, hopefully, but you I’m sure it will be some others. But I think that would be the bet is that there really is a benefit to the compounding firm. And so when we think about how we expand, whether that’s around practice areas or client types or complexity of work.

The ideal would be that, you know, a law firm partner joins us, brings their book of business, but really like this is the best place for them to work with their clients because of the whole system beneath them that we can offer.

RT (13:36.908)
Yeah, that makes a lot of sense. Yeah, exactly. You’re moving to real estate, employment law, etc. You just keep moving out. And maybe you still initially steer away from the billion dollar contentious issue, but you just get you just start munching away at that bottom third of work and you just just but you you go laterally across the market. Yeah, that could definitely work. Definitely work. Let’s talk about what you’re doing today. There’s a couple of announcements. Just talk us through the benchmark just briefly.

and also the new research group on agents.

Ryan Daniels (14:09.406)
Yeah, no, I’m excited to share more about this. We have been working for the last few months on a specific kind of segment of research all around kind of getting agents really great at what what I guess we would call exercising legal judgment. And so when you think about where AI and the you there’s there’s there’s some also great benchmarks that already exist. And so we’re kind of just contributing to the open corpus, which which is you know what we’re really excited for. When you think about

The kind of moment that AI had in coding over the last few years, it’s really fueled by these verifiable rewards, which is to say there’s a right answer for some coding exercise or math problem. And anybody can look at it and say that was the right answer, that was the wrong answer. And so can get these feedback loops for you know for like the kind of reinforcement learning for models that make them just exceptionally great at coding. and the problem that we keep running into is in a lot of the legal tasks that we’re doing, it’s heavily judgment-based. In other words, there is no right answer for.

how you push back on a counterparty, you know, redlining some term for you. and

You know, when we try to kind of do like a reinforcement learning loops, you get really stuck because there are three or four great answers. And and honestly, three or four lawyers on our team would reasonably disagree on like what was the best tact here. Should you have been really aggressive? Should you have been very conciliatory? should you have just accepted it? Should you have pushed really hard? And the end result is like opaque and it’s kind of a moving game. This is just we’re talking about contract read, right? So this is like simple contracts, not even that complex, and yet it’s still quite hard to kind of think through like

What is the best answer? And ultimately, what you’re paying a great lawyer for is their judgment, is for them to tell you, look, we should really push once more here. I I just knowing the counterparty, knowing what I know about the industry, I think it’s I think it’s reasonable to push one more time for like, you know, a higher liability cap or to indemnify for this other term. And what we kept seeing was, you know, until agents can do that, we can trust their reasoning and the judgment calls they’re making.

Ryan Daniels (16:11.86)
You know, we’re gonna we’re per like we this law firm, we’re fundamentally capped at needing lawyers to give more and more input and and to weigh in more and to spend more time on things. And we get paid, we don’t have billable hours. So like we are very incentive to try to push as much work as we can to agents. And this is just like a fundamental limitation of the models and of the architecture and the way they work. And so we wanted to start investing resources into like research around improving the models for this kind of judgment context, this judgment basis.

So we’re doing a few things. One is I I g one one is, I guess, this intelligence group we’re calling Crossway Intelligence. We’ve been quietly working on this for a few months now, but we have a team of mostly ML engineers working with lawyers who have kind of a technical background to just push the limits of what these agents can be doing. And the other is publishing today, and we’ll continue to contribute to it, a benchmark that shows it’s the first benchmark that we’re aware of that shows multi-step negotiations. So tracking and measuring.

RT (16:40.397)
Ciao.

Ryan Daniels (17:10.296)
LLMs at each step of a negotiation and how do they mimic and mirror lawyers and where lawyers have consensus on what’s the best way to push back and lawyers have very little consensus on what’s the best way to respond to each other. And in doing so we can see like how are agents or models able to negotiate with with one another over time and make the right judgment calls.

RT (17:28.109)
What I like about this is, as you say, really is in the judgment layer. It’s not about just basic accuracy. know, like, you know, did you read the document correctly or did you use the playbook correctly to draw out the correct language? It’s about persona. Right. It’s really getting very human, which then leads me to the next point, which I was going to ask, is, could you, and I know some people do this already to some degree, you you can sort of like…

like move the gauge up and down. Can you set it to become more aggressive? So you’re working for a particularly famously aggressive real estate company that argues every point until the other party collapses. And they go, look, we don’t care. We’ll just set your agents to maximum force, right? You know, set them to kill, right? And you go, okay, all right. You may get some pushback on that, but we’ll set the agents to maximum. And then you get everyone to quite relax and go, look, we won’t, just want the deal then.

quickly. That’s our primary objective. Obviously, we’re not foolish, but as long as it’s within this scope, we’re happy. I mean, is that kind of the way to go?

Ryan Daniels (18:31.816)
I think I think yeah, like I think what you start to see is, and this is why, you know, I’m so glad that we have this law firm and we have lawyers who are for the most part like, you know, we have like no super senior partners, but more senior lawyers, right? ‘Cause I think we need a bit more experience to kind of keep giving feedback to the models. And what you see with our lawyers is there is very little uniformity in how they approach things. They’re all great, right? And some are some are very conciliatory and it depends on the client, depends on the context, depends on the real. Right.

RT (18:56.215)
to pass that really inside.

Ryan Daniels (18:58.752)
You know, and so when we start and it was like three lawyers, like me and two other lawyers, like we kind of assume like let’s have a streamlined unified Crosby voice and Crosby way of working. And when you work with us, we know what you get. And like it turns out that that’s not really what clients want. They want to know who their lawyer is. And is this person like, you know, a bulldog or are they really, you know, you know, kind of

RT (19:18.797)
And then there’s an interesting point which is then it becomes self-selecting. So certain clients will seek out certain lawyers who mirror, maybe consciously or unconsciously, their own desired outputs.

Ryan Daniels (19:31.753)
Right. And that’s exactly right. And so we always thought, like, let’s just look at the outcomes and kind of work backwards and see like what was the most effective way to have this agent negotiate, right? And like work backwards. But it’s every negotiation’s a little different. And it depends as much on the counterparty, you know, when you’re negotiating against Walmart or something, right? Like like, you know, you th doesn’t matter how aggressive you get, like this like you’re c you’re kinda limited from the jump. And when you negotiate against a smaller startup,

you know, there are there are concessions you can get and you just have to know how to kind of ask for those and what the right ones are. So there’s some data that you can inform and be a smarter negotiator, but ultimately kind of your approach matters. And so what we’re thinking more about doing here is, you know, how can we just start to hill climb agents on replicating the judgment and persona and intuitions of specific lawyers at the firm? And if we can build agents that sort of mimic me and my style or mimic you and your style, that just becomes really scalable.

And we can just have leverage for our lawyers that feels like the you know, the kind of agents that they that that that like

RT (20:30.317)
You see, you more like avatars and you have like, it’s almost, I don’t know if you’re into Dungeons and Dragons or if you were when you were a kid, it’s almost like you create player characters, right? And you pick one and you get like, you know, get like a little, you know, little thing with a little bubbles for numbers in and stuff. This is, this is Thrud. Thrud is a barbarian, you know, he, he fights every single claws, right?

Ryan Daniels (20:46.506)
Totally.

Ryan Daniels (20:53.64)
Yeah, or it’s like you know, like if you were to think about, you know, designing your ideal like like like basketball character on some of the basketball video games, right? And you could like have all your attributes. But I think like, you know, and so so but you can and there are should ways to tweak that, right?

RT (21:04.685)
Serious point, it? You effectively, as agents effectively, I just glorified motors stuck on top of a bunch of workflows, which in data that we’re allowed to access. But as we get more sophisticated, I guess eventually we will get to these avatar type scenarios, which are going to be super agents with judgment. Maybe like agents with little brains attached.

Ryan Daniels (21:29.504)
I mean, I think when we just zoomed out and looked at like all of our experience at and you know, most of the lawyers here came from from bigger firm big law firms in New York, you know, our first few years, the first few years you have with a partner, you know, like and you have a few partners you report to and they each drill into their way of working. And like there’s su like what like w like somebody joked here, one partner they worked for, you never used the word shall always use the word will in drafting. It was like a brightline rule. And the other partner was exactly the opposite. they were like I just you know

And so you you you you kind of have to conform your style and just to their their taste. And you know, ultimately some things are trivial like that. But other things, there there should be a way to build agents where it really is an extension of a senior lawyer and all of the loss that happens between trying to teach a human over and over again your instincts no longer exist. And so that’s kind of what I think we’re working towards.

RT (22:19.341)
But I mean, that’s really interesting because I think a lot of people, particularly at traditional law firms, might just be like, oh yeah, these new mods, these AI first law firms, they’re they’re characterless factories churning out contracts. But actually, really, in some ways, it’s actually quite the opposite in that you’re you’re sort of tuning in very, very precisely into the character of your lawyers, much as perhaps in the same way that they do. Or in some ways, perhaps even you could, I mean, it’s an interesting thing, isn’t it?

You can actually deviate from the big law norms by saying, there is no corporate identity here. Norm or Crosby or whoever is a collection of personalities, 80 % of which are automated and you pick which one you want.

Ryan Daniels (23:04.468)
You know, I think it’s a really good insight. I would say when we started, our intuition was, let’s just be as sort of unified as possible and so you know what you get with Crosby. And

RT (23:17.207)
Yeah. Which is what the outside world thinks. They’re just like, yeah, you go to Crosby, you check them an NDA, it’ll be the same as all the other NDAs they do. It’s just ching ching ching ching.

Ryan Daniels (23:28.286)
Right, right. Which is which is which was I think like a which I think is, you know, I think kind of how some of these like legal process outsourcing firms work. You kind of build your playbook and your parameters. The truth is, you know, when you look at where all the value compounds for law firms, you know, these you know, top hundred law firms account for like thirty, forty percent of all legal spend in the US, it is for the judgment preferences and taste of these really experienced lawyers and it’s something that you can mimic and replicate.

RT (23:37.239)
Yep,

Ryan Daniels (23:54.763)
you know, brilliantly with agents and you and you really can make them a little bit I we’re to be clear, we’re not there today, but you there’s a there’s a there’s a clear line to making these agents actually be able to ingest more information than a typical lawyer and be a little bit smarter just just because they have more information available to them. And so why shouldn’t that happen? And you know, why shouldn’t all the taste and intuition and just like you know kind of pattern matching that a senior lawyer brings be able to be offloaded to agents. I think our ideal is a law firm where

you know, a senior attorney, a partner works here and feels like they have this remarkable team that is working for them that kind of just has a mind meld from the jump. And it should be actually like and you know, I think again, in order for us to be a serious contender in the legal industry, we need to be a place that law firm partners really, really want to work. And I think

RT (24:42.207)
Exactly. just about to say that you can’t, if you come across as just look, we’re just like a robotic sweatshop and we’re the lowest common denominator and we make our money just through volume. Well, any, any really good lawyer will be like, well, okay, I don’t mind co-owning this with you because hey, I’m going to make some money, I’m no, there’s no way I’m going down onto that factory floor. But if you take, if you flip the script and just say, actually, no, it’s not like that at all. It’s actually going to accentuate you. We’re actually going to make clones of you.

Ryan Daniels (25:09.781)
Yes.

Yeah, I mean I think by the way, it’s it’s probably a good business, right? To just say, like, here’s the unified voice. We tweak it a little bit to our clients’ preferences, and then it just kind of goes. And and and I would say that’s pretty much where we are today, right? Like we have a good amount of customization on a per client basis, such that it blurs the line between what would be your external lawyer and what might be internal, because we just remember a lot about you. And that’s like a delightful experience. But you know, as we do more and more sophisticated work, and I think that’s something we’ll offer, but as we do more sophisticated work.

We need to be able and we we we take a long view on the models and we know they will get to a place where they can be like a wonderful companion and you know sort of junior associate for great lawyers. And so let’s take that long view and build that law firm. And that means, you know, really being able to tweak it to the way they work.

RT (25:59.598)
Yeah, I think, you know, let us think about, know, like when you go to a restaurant, okay, you go for the food, but you go for the ambiance. You know, if you keep going to a certain restaurant, you get some of the waiters and everybody, right? And you’re starting to buy the personality, right? You know what mean? And I think that’s the same for any high value service industry. So maybe in a funny way, actually, new mods actually have an opportunity there because you can…

Ryan Daniels (26:14.686)
Yes.

Ryan Daniels (26:20.713)
Yeah.

RT (26:28.533)
You can, basically, if you can turn the personality of a great lawyer into a piece of software, then you can multiply that piece of software infinitely. Because as someone explains to me many, many, many years ago, we were talking about the difference between making a flint in the Neolithic era and making a piece of software. So it could take three months to make a great flint axe. And once you’ve made that one, it made no difference. Even though the person became better making that flint tool, the only way to make another one was to go and find another grabbing piece of rock.

and go at it again. Piece of software, once you build it once, infinitely reproducible.

Ryan Daniels (27:05.406)
No, I I so like you I think you’ve summarized it better than than I possibly can. And I do think every law firm will need to do this. I I think we just happen to be a little earlier. But the value is when we think about our long term value, is being able to offer more access to excellent lawyers who are constrained by the fact that they are one person with a couple of associates they have to oversee and manage. But, you know, what if you could have access to like

RT (27:15.341)
Mmm.

Ryan Daniels (27:34.367)
you know, these sort of agents that report up to David Boyce and like encode, you know, his judgment and, you know, way of working and thinking and and, you know, and I think that becomes extremely valuable. And so we constantly talk internally about like how do we encode? How do we capture all the judgment of our lawyers, and and and and scale it and replicate that. And again, I think the experience of a lawyer when you can offer that for them is like, this is exhilarating. Like I I’m able to, you know, do so much more work and

see so many more clients and have all this exposure because I have these agents that I’m overseeing. And it’s a delightful experience. And I think so long as a law firm is structured to be very incentive to do that and to offload as much of the lawyer’s judgment work to agents, that lawyer becomes more valuable. But the business model has to be able to incap incorporate that, right? So there’s no, you know, no billable hours and you know, the profits that we make every year get reinvested back into RD. And I think that’s the key point is, you know, with the MSO model is

There’s a real incentive to keep reinvesting back into to new products because the models keep getting better.

RT (28:38.093)
Yeah, yeah, yeah. And as another company was telling me recently, mean, the key point to them was also that reinvest the profit because you will then improve the workflow, which actually increases your profit margin. So you might sell off your profit margin and then next year it’s 50 and so on because you’re just smoothing off the edges. let’s talk about business in general, right? So for 50, 60 years, the Big Law model has reigned supreme, very, very large amounts of pyramid.

Ryan Daniels (28:50.76)
Exactly. Right.

RT (29:08.237)
leverage time, et cetera. And, you know, as the co-founder of the very, very well known legal type platform, you know, said to me, look, you know, we, we’ve got to stop giving big law such a hard time about billable hour. Because if you make things completely with people, how else are going to charge them out? I said, well, that’s fair enough. But then once you introduce AI, well, then how do you do it? And I think this is, this is the thing. And you, you luckily have.

Ryan Daniels (29:28.16)
Mm.

RT (29:38.518)
you’ve just dodged our problem. Whereas the big law firms, think quite earnestly are trying to solve it, but they’re finding it extremely difficult because they are wedded so deeply to this infrastructure, this kind of time equals value, high leverage, bodies in a room, you know, type approach. It’s very, very, very hard for them to flip. And they don’t want to lose their clients. It’s like going down the freeway at 100 miles an hour, you know, rebuilding the car as you go.

You know, the cliche about trying to change a wheel on the car while it’s traveling down the You guys really, and you and others who using similar model, you don’t have any of that baggage. So how far do you reckon, not just you, but all the other new mods, how big a chunk of the market do reckon you guys could grow into?

Ryan Daniels (30:30.314)
Look, I think there’s a lot of stuff we’re figuring out, right? Like we didn’t know the judgment constraints of AI and think we’d be doing a benchmark two years into it. We I mean we there’s just a lot we’re figuring out. The one constant that we really, really understood from the very beginning was the business model has to have aligned incentives to take advantage of technology that can actually do work that only skilled humans could do before. If if that’s not there.

you’ll you’ll you’ll you just you know you might get you know some percentage of the way 10, 20, 30, but you won’t take it to ninety, a hundred. And that was our starting proposition. You know, I obviously I’ve done a ton of research, you know, especially in the early days on the history of the buildable hour, which is it’s actually not that old. Like it’s it started really in the sixties and seventies as legal work became more complex. And it makes conceptual sense for some frontier work that ex ante is just impossible to price, I guess.

RT (31:16.909)
Yeah.

Ryan Daniels (31:29.822)
But I think a few things. I think one, I think you’re right. I think, you know, you know, I think I think losses hurt more than foregone gains, right? We just know this. It’s very hard to disrupt yourself. We’ve seen this time and time again in history. I think there is

I think a very hard prospect for law firms, if we’re being very honest about it, is law firm partners making millions of dollars a year in salary. That is just the profits of those law firms being distributed out each year and not being reinvested. And when you look at any technology company, all the profits get reinvested, and those millions of dollars are just equity in the business, right? It’s not cash salary. And that’s how you’re able to make long-term investments.

You know, we looked at the top hundred law firms last year, made seventy billion dollars in profit total, which is the same amount of money that Google invests in R and D each year. And it’s just an astonishing sum of money. And so, you know, I think this business model made perfect sense in an era where

You were providing human capital as a service, and the winners were able to attract, train, and retain the h the best capital, the best people.

RT (32:42.253)
Exactly. So the more money you retain, well, the more money you can actually show to the market because like the Amalor 100 table wasn’t just a breakthrough in that it provided transparency. It became a self-reinforcing advertisement for that model because young lawyers or young wannabe lawyers saw the numbers and went, geez, are you kidding me? You can make that much money. I’m going to Cravath, right? Or I’d like to, you know? And that became a self-fulfilling prophecy.

everybody up like a convection cycle.

Ryan Daniels (33:14.942)
I think and and and I mean I mean listen, like I remember looking at those numbers when I was in law school and kind of being amazed. I think, you know, if the you know, if the the billable hour rates are increasing greater than inflation year over year, over and over again now. I think it’s like we’re going on eight, nine years. Something’s something’s up, right? Like either the supply isn’t growing fast enough with demand, like some like it’s just it’s it’s it’s a broken market. And

I I wanna be careful and say like I you I don’t I I I have nothing but respect for the partners I worked for, you know, and who are like truly brilliant attorneys. And you know, I under and and the market’s

RT (33:49.707)
Yeah, let’s face it, mean, to be fair to Big Law, there’s a huge amount of demand for their services.

Ryan Daniels (33:56.713)
Right. And the market is the market. So, you know, like I think like a lawyer a couple months ago said, you know, what’s my bill of all hour? It’s what people are willing to pay, and they’re willing to pay for three thousand dollars an hour. So I understand it. I’ll just say I’m skeptical that it’s doing right by society when prices are going up this much. And, you know, law is like a good that I think more and more people should have access to and have access to great services. I think by you know, we we lose money on work all the time, right? And it’s a it’s a terrifying s prospect for us because we priced it

improperly. But I’d rather us take on that risk than our clients. I think we do right by them in that instance. And there are a lot of lawyers who hear me say that. They just think it’s like a light turns on. They’re like, I’ve always felt that way. Like we have a duty for clients. I want to, you know, this is this is the model that I want to build. So, you know, I think I think many big law firms will adapt. They’ll have no choice. but it will be really hard. And, you know, luckily we just we kind of start from scratch with it.

RT (34:51.501)
Yeah, yeah. mean, my long-term theory is in about 10, 11 years from now. I’ve got the date 2037 lodged in my mind. It came to me in a flash of inspiration, bit like how I started artificial law in 2016. I don’t know how, and there were several steps. Even in 2016, I thought we were kind of going to get to, but I had no idea that LLMs were going to be the route. I thought natural language processing was just going to get better and better, and of course, it never did.

Ryan Daniels (35:04.64)
Yep.

RT (35:20.299)
Anyway, I kind of feel that somehow you’re going to end up meeting in the middle. I think some of the most progressive law firms are going to become more like new mods and the most ambitious new mods will become a bit more like law firms in the traditional sense in that there’ll be multi-practice groups and they’ll actually go up the value tree so far, right? And then somehow there’s going to be a kind of merger in the middle. And I think that maybe will be the birth of a completely, completely new model, even more radical than what you’ve already created.

Bye.

Ryan Daniels (35:50.089)
Listen, you know, the even the idea, even talking openly about like maybe wanting to compete with that ammo hundred frame is kind of like it’s bizarre. But you know, like we’re you know, like we’re just figuring out we have some great lawyers here, we’re in New York, and like I I feel proud of the work we’re doing, but like, you know, these law firms are like they make they make the economy function. Like, so I have no illusions. But I think if we can if we can, you know, you know, make the elephant dance, right? If like if we can, if we can just prove that there is a model that works, that is

RT (35:55.467)
Yeah, it’s mad.

Hmm.

Ryan Daniels (36:19.422)
Better for the legal ecosystem, is a you know is better for clients, then we’re doing our job right. And like I I I like you can probably tell, like I passionately believe that the role that lawyers play, especially in US society, is essential. Most Americans don’t have access to legal services, most American businesses don’t have access to the right services. And so, like, and and now we have a technology that already can change that. And so the question is how do we drive adoption as fast and intensely as possible? And and so that’s all we’re here to do.

And I’m glad that the industry is starting to take notice. And we’re not the only, you know, sort of neo firm doing this, but you know, I think I wanna be loud about it and you know, would love to grab some lawyers from some of these firms to make them notice like there’s something going on here. I think, you know, what what Kirkland announced a few weeks ago is an amazing step in the right direction for starting to invest money in yeah.

RT (37:06.573)
Yeah, it’s a real signal. Real signal. Yeah. mean, as I said to several people, it could be an involuntary Trojan horse. They’ve kind of Trojan horse themselves in that they were very, very keen to show their clients they were taking AI seriously and went pretty heavily into it. The problem is, that the clients are now watching them very closely and in about six months they’ll be like, all right, so you’re kind of radically changing whole swathes of your practice groups and they’re going to have to do something.

Right? Otherwise it’s going tremendously embarrassing when people are like, there’s going to be write ups, right? It’s going to be write ups. And like when we say, well, yeah, we haven’t actually changed anything. It’s going to be, or we did change some things, but really if you actually dig into the numbers, it’s cosmetic. I mean, that’s not going to wash, right?

Ryan Daniels (37:51.743)
Yeah, I I’m I’m fascinating. I mean it’s it’s it’s it’s a lot of money to spend on R and D, which and you know, and and so it’s terrific and so it’s like I’m I’m fascinated to see where they end up. And you know

RT (38:02.445)
Well, kudos to them for doing it because I mean, you know, they’re very, very, very clever people, quite obviously. So I’m sure they thought all of these things through already before we even took those steps. But I mean, it’s and if I do mean it for real, I hope they do. It’s going to send such a signal to everybody else.

Ryan Daniels (38:18.452)
I I mean I think even now it’s a shot across the bow that like a lot you know, it’s I think it comes out to about a percentage, it’s ’cause it’s over four or five years. It’s about a hundred million dollars a year, it’s a percentage of their overall revenue. So

RT (38:28.365)
Which is like, let’s have a sandwich budget, right?

Ryan Daniels (38:32.36)
Right. Right. Right. So you think about compared to like, you know, any tech company where it’s like thirty percent of their revenue each year is already. So okay, so but but that’s still a lot of money to spend when you just think about like some of the AI legal companies and how much they’ve raised, this is like you know, already about as much as some of the big ones have raised. It’s it’s it would be fascinating to see how that goes into R and D.

RT (38:52.621)
Yeah, yeah, yeah. No, totally. Well, let’s just briefly talk about money and numbers and we’ll end. So you got about 85 million dollars recently in total. that a big round?

Ryan Daniels (39:03.264)
That’s our that’s the total amount raised, yeah. So between our seed A and series B.

RT (39:08.055)
Gotcha. you’ve got, you know, if you go on Tiles for Shoreline, you can see a list of all the investors and really, really big names. How many people are you? And I know actually, I mean, obviously one of the ironies of asking that question is of course, is that in your kind of business, just having lots of bodies doesn’t necessarily mean that’s a good thing, right? You actually want to have as few as possible. You know, you have a $1 billion startup kind of thing with one guy. How many staff do you have?

Ryan Daniels (39:26.24)
A hundred.

Ryan Daniels (39:33.056)
Mm-hmm.

Ryan Daniels (39:37.44)
So we have about we have about forty five attorneys now, almost I guess probably fifty as of this week. and then about another forty folks kind of between the engineering and and operations teams. So we’re getting close to a hundred. but to your point, you know, our our core metric, especially on the law firm side, is the kind of productivity per lawyer. Like this is every week in our all hands, this is what we’re really being.

RT (39:52.098)
Yep.

RT (40:03.189)
Yeah, I know, bet your productivity per lawyer for you guys compared to an equivalent firm doing that work or an equivalent practice group doing that work, you must be so many times higher.

Ryan Daniels (40:13.46)
I you know, I I it’s I I we don’t it’s hard to get benchmarks for law firms, especially because there’s like so many models of leverage between junior and mid-level senior associates. But yeah, I think, you know, out of the gate, right, you have basically agents more or less doing a first pass at most work. We’re just starting to have agents that, you know, can work autonomously doing basically paralegal work where the lawyer doesn’t to see it, which is what we’ve always wanted to get to. And, you know, answering simple questions, cleaning a document, updating a signature form or something. The client could just say, Can you do Paralegal can can can do that. But

Yeah, we just are seeing that, you know, tasks that took hours can take ten, twenty minutes for a lawyer and it’s magical when that happens. And and then also lawyers that would have had to search for a long time for some, you know, small detail in in you know in a in a former document, it can just come up in minutes. Like these are the small things that are really wonderful when they start happening.

RT (40:59.437)
Yeah, fantastic, fantastic. Well, look, thank you so much. And we really look forward to hearing much more about your growth in the future. Thank you very much.

Ryan Daniels (41:10.494)
Richard, it’s it’s an honor to be on the show. It’s like full circle from from having gone to conferences before before we started the company. So this is this is so exciting.

RT (41:19.32)
Thank you.