Inside How Crosby Is Automating the $1 Trillion Legal Industry | Ryan Daniels of Crosby

Law is a trillion dollar industry every year and that's what we spend on lawyers globally. We focus on commercial

globally. We focus on commercial agreements for mostly fast growing tech companies. These are MSAs, DPAs, NDAs.

companies. These are MSAs, DPAs, NDAs.

[music] Any contract that a business needs to sell it software, we handle those. I

actually spent time in law school working for a court in East Africa for a judge. You could see the rigor of the

judge. You could see the rigor of the judicial opinions there was just nothing like it is here. There's something great about the fact that [music] it takes so many months to get a case through in the US, but you get to a brilliant result.

is a thousand lawyers. That's bigger

than top 50 law [music] firms. They have a huge law firm within Google. Law firms

that are selling legal services to a bunch of clients have a lot of restrictions on how they can organize [music] themselves. Quality in legal

[music] themselves. Quality in legal work is this really beguiling issue. If

you ask somebody, why do you pay that lawyer $2,000 an hour instead of that lawyer $1,000 an hour? Is he really two times better? It's probably impossible

times better? It's probably impossible to explain why.

>> Ryan Daniels, co-founder and CEO of Crosby Law. Welcome.

Crosby Law. Welcome.

>> It's great to be here, Jason. This is so fun.

>> I am so pumped to have you. Uh, some big announcements you just had recently with your fund raise and I noticed a name on there, Sequoia, which has a number of investments in the legal tech world.

What's the thesis here? Like what are they going after behind all these legal investments? Well, I think uh something

investments? Well, I think uh something I've started to say a lot is I think we're in the golden age of legal innovation. And what I mean by that is,

innovation. And what I mean by that is, you know, law is a trillion dollar industry every year, that's what we spend on lawyers globally. And of all the pillars of our society and economy,

like healthcare and accounting and insurance and education and and real estate, law is by far the most manual to this day because of the nature of the work. And nothing really changed about

work. And nothing really changed about that until LLM came out three years ago.

And so I think for the first time the way that law is practiced and the ability for people to access it is going to really change because technology can finally make law slightly more automated. And so if you have a trillion

automated. And so if you have a trillion dollars spent on almost 90 plus% human labor and that can be automated. I think

Sequoia like many VC firms are realizing this is an enormous opportunity to finally um build really really big software companies. And so what I think

software companies. And so what I think Sequoia understands particularly well that that trillion dollars spent on legal and legal as a category is as diverse as transportation. So there's

bicycles, there's Ubers, there's airplanes. And so law is just there's so

airplanes. And so law is just there's so many different things. There's

litigation, there's trust in estates, there's corporate law, there's commercial law. I mean it's really

commercial law. I mean it's really really diverse. And so I think they've

really diverse. And so I think they've been um pretty clever to see all the different verticals and understand there's um very very big software companies to be built in each of those verticals and and probably more than one

company in each one.

>> I'm curious what your mental model is of the landscape of legal tech. Like when I think about it, I think about there's in-house counsel, there are out

outsourced GCs or outsource counsel, there's different types of laws you were talking about, like what's your mental model specifically when it comes to the legal tech world.

>> So I think again I think I would go just when you break down that trillion dollars that we spend on lawyers every year, which is it's just that number is kind of hard to get your head around and I think it's easy to trivialize.

um this is bigger than the GDP of the majority of countries in the world. You

can just see like where that money is going, right? So a huge amount, a

going, right? So a huge amount, a disproportionate amount goes to large companies and funding litigation or M&A or corporate work that they're doing um just because they have the ability to

pay. I think um another big chunk of

pay. I think um another big chunk of that at least in the US goes towards plaintiff's law. of this is, you know,

plaintiff's law. of this is, you know, if you're wrongfully fired or terminated, if you have a health issue, if you get in a car accident, um if there's a disability insurance claim, I think that will probably go up more. So,

there's like a huge um element of consumer law. It's just people are

consumer law. It's just people are priced out of it. And then there's this really long sort of tale of all the things in between like small businesses and all the needs they have. I I I you know, wish I had more data on actually

that breakdown of the trillion dollars of spend in the US. I can tell you we spend around $400 billion dollars on on legal services. So 40% of the global

legal services. So 40% of the global spend on lawyers is is just in the US which is pretty amazing and about half of that goes to about 100 law firms which is pretty wild too right so it's really topheavy but then then the the

other half goes to um about 100 thousand law firms or more so it's a really long tail that's divided up >> that's wild um where does Crosby fit into this landscape and what is Crosby for those of the people listening that

don't know what it is >> yeah so Crosby is an AI first law firm that is focus focused on automating humanto human negotiation.

What that means is a contract is fundamentally just two humans negotiating on a handful of terms, but they're really collaborative and they're actually trying to get a deal closed, right? And so this is the price of the

right? And so this is the price of the contract, how much money is being exchanged, and there's all these other small terms like limitations of liability or different warranties or assignments of IP, things that don't make a lot of sense, but there's a

dollar value to each of those. And when

you really understand that every contract, every transaction between two parties where there's a mutually beneficial exchange of value, where dollars are being trading hands, there is going to be a contract that just

makes that run.

Then you start to see that anything from like a two-page offer letter to an 100page merge agreement is just a negotiation that you can automate if you understand both parties sides. The only

way that we could automate this properly was to build our own law firm. That way,

we're not just a software solution used by some lawyers. We do the entire negotiation. Um, we receive a contract,

negotiation. Um, we receive a contract, we make edits to it, we predict what the counterparty might say or might not say, and then we send it back.

>> Can you unpack that a little bit more?

So, maybe walk me through a specific example of a contract that you guys are reviewing and what you're doing for your customers.

>> Yeah. So, that's a very abstract explanation of Crosby where we're practical today is we focus on commercial agreements for mostly fast

growing tech companies. These are MSAs, DPAs, NDAs. Any contract that a business

DPAs, NDAs. Any contract that a business needs to sell a software um and you know any tech company will be very very familiar with these and how painful they are to negotiate. We handle those and

you work with us like any other law firm. You slack us, you email us your

firm. You slack us, you email us your document. You give us a few points.

document. You give us a few points.

Typically a salesperson or an AE will just email us or slack us. Sometimes

their lawyer and we get it back to you in an hour or two max. And we have a lawyer on our team who's a bar attorney all based here in the States, most in New York. I'm reviewing it for you. So,

New York. I'm reviewing it for you. So,

it's safe. We have like a human judgment, but um we have a lot of AI workflows we've built in to make that incredibly fast no matter what time of day.

>> That's incredible. And so, for years, there's been this market called I think it's the LPO market, right? Yeah. Is it

legal processing?

>> Legal process outsourcing.

>> Outsourcing, which historically has been a market with Fortune 500s, Fortune 1000s like outsourcing this stuff even overseas right?

>> Yeah. And so did that inspire the business and the model that you guys went after here?

>> It was it was important to our thinking.

We didn't want to build an LPO, but you might recall I actually spent time in India when we were thinking about starting the company because I kept hearing rumors of large corporations in

the US who had big offshore teams predominantly in India but also in Eastern Europe that were doing some of the more formulaic legal tasks. And that

was very strange to me because in law school you learn a lot about how you have to be a barred attorney in the US to practice legal work. But it turns out that there's an enormous amount of legal work that you can do according to some

rules. Um it's actually pretty

rules. Um it's actually pretty formulaic. And you can make a really

formulaic. And you can make a really good case that if you are an offshore attorney just applying some rules to a contract, if you see 30 change it to 50, if you see Delaware change it to New York, then that's not legal work. Um

it's not the practice of law. So I

actually spent time in India. I met a lot of these companies. They're they're

big. This is like a 14 billion industry.

It's growing 30% year-over-year. So

that's actually bigger or about the same size of the market for all legal software. We didn't want to build an

software. We didn't want to build an LPO, but we understood viscerally that companies want a faster, cheaper, and reliable legal service rather than some AI platform that plugs into their word

document. Some some lawyers want that

document. Some some lawyers want that and they're very helpful, but most companies just want to throw their legal work at someone, know it's in good hands, know it's taken care of, know it's fast, and then get it back really quickly. And that was I think one of our

quickly. And that was I think one of our light bulbs that went off when when I was in India.

>> It's kind of interesting because you guys decided ultimately to follow what feels like is the current workflow for us today. Like I get a contract or I get

us today. Like I get a contract or I get a term sheet and I send it over to my council and then you're just sending them back the red lines, right? What I

think we do well at Crosby is prioritize empathy and sometimes too much where we're trying to do things that are just not automatable or automatable or scalable because we're just like so viscerally feeling what the clients are

feeling. But one of those things and

feeling. But one of those things and again this is a product we use Crosby internally for all of our documents. So

we know the feeling of like you get a document you're super stressed you have a bunch of emails and you're like I just want someone to look at it tell me what I can and can't accept. Just give me the thing that I can just send. I don't even want to open it. I don't want to download it and look through the margins and answer questions my lawyer has. I

just want to take it know it's ready to go and send it back to the counterparty.

That experience is so delightful of just like I don't have to deal with this anymore. I've taken some mental burden

anymore. I've taken some mental burden off of my plate and just handed it to someone else and they just deal with it.

I think software and services but particularly software with AI today at its finest does that you just you offload some mental burden and some work burden to um your vendors to your

software it's pretty fascinating is how do you think about the pros and cons of that interaction being like it's still an email right is there a clientf facing user interface at all or is it

specifically email based >> yeah so it's actually mostly Slack we we do email most of our clients prefer Slack and so we've actually invested a bought in a pretty sophisticated Slack integration. So, we have a bot. You tag

integration. So, we have a bot. You tag

it. It It can give you updates on what your document is. Is it being escalated for further review? Um, for some clients, we're piloting giving complexity scores. Hey, this is going to

complexity scores. Hey, this is going to take x minutes. This is how um how hard we think this one is. It's, you know, kind of hit and miss with AI doing that.

Um, and then, um, we can send back some questions. Hey, we have a few more

questions. Hey, we have a few more clarifying questions on this document, like how big is the counterparty? Is

this a regulated contract? Whatever. and

get that all on Slack and we send it right back to you. And you know, I think that there's a bit of a SAS fatigue of like having all these platforms just to

upload a document to. And um I'm pretty bullish on Slack as kind of a super app of the future because you just have so much you can build on top of it. And you

don't need to click buttons anymore when you can just with natural language say please do X, Y, and Z for me. Um having

said that, Slack is a very limited interface, right? You can just speak a

interface, right? You can just speak a natural language. There's a lot that if

natural language. There's a lot that if you could just have some buttons or some graphs or some charts, you could really show great things to the clients. So,

we've just started launching in beta for some of our bigger clients. Um, this

really elegant dashboard that just shows you where are all of your contracts, where are all your negotiations, where are you stuck? And for salespeople, this is amazing because, you know, you might have not heard back from a party for 90 days and the deal is still live, but

it's just stuck in their procurement team. They're a big Fortune 500. And we

team. They're a big Fortune 500. And we

can tell you, look guys, we sent this back 45 days ago. This is an ongoing negotiation. this deal will not get

negotiation. this deal will not get stuck. The legal flags here are not

stuck. The legal flags here are not going to be deal breakers and you should go ping them. And so we give you a really nice dashboard. The best part about this dashboard is it's just through Slack. You don't have to upload

through Slack. You don't have to upload documents. You don't have to tag them.

documents. You don't have to tag them.

You don't have to categorize them and fill out fields. You just Slack as a document and we extract all the relevant data points of it. Who are the counterparties? What's the deal size?

counterparties? What's the deal size?

What's the urgency? And you have a beautiful chart with all of information there. I think more and more companies

there. I think more and more companies are going to start building, you know, sophisticated integrations here. I think

Ramp where my co-founder was was an early employee is a great example of they have wonderful Slack notifications to just do the most essential things.

Approve expenses, approve payments. Why

should I have to open the Ramp app? They

don't really prioritize clicks in their app. They just prioritize getting your

app. They just prioritize getting your workflows done faster. And so it's it's an elegant way to do this. It's really

interesting to think about Slack as the central hub moving forward. But it

actually makes a lot of sense, right? If

you can have a lot of data living in that system and it can then go out and work with a Crosby or I think there's a company that you guys actually work with or uses Crosby that collects all of the data from all the engineers and

generates the reports on what is the productivity for your engineers and then plugs that into Slack. It kind of makes it that human the loop like ecosystem

for any employees of the future who want to be using agents outside of uh their core workflows today.

>> Yeah. So that's right. It's a great So, I think we're seeing this more and more.

And what's really cool about Crosby is the companies that get the most value from us are the fastest growing companies, most of them in history, right? Like Cursor was one of our

right? Like Cursor was one of our earliest clients, I think, literally the fastest growing company in history. And

so what that means is we seem to get a bit of a vantage point into the companies of the future because they're the ones that are banging on our door asking us for access to faster contracts. So Kavon's company,

contracts. So Kavon's company, Macroscope, is what you're talking about. Um, they allow you to have a

about. Um, they allow you to have a really sleek interface all through Slack. another one of our clients

Slack. another one of our clients console is sort of like a managed IT solution that integrates to you know let's say you're trying to create a new um Google account for a new employee you can just slack it and it figures out

kind of where to route your request throughout all of your applications and they've built a really elegant Slack interface so all to say we get to see the companies of the future and a lot of those companies of the future are building through Slack by the cursor has

a Slack integration you can slack cursor and ask it to code something for you so it's just um it's so obvious because it's so easy and and I and I do think there's an enormous amount of time spent every day toggling through different

applications and getting stuck and and slowed down.

>> So, what I'm hearing is we should buy Salesforce stock, right? Because like

they bought it, man, >> this might be the lifesaver for Beni off. Uh it might not actually be agents.

off. Uh it might not actually be agents.

It might just be Slack. That was the best acquisition ever.

>> Maybe it's just agents through Slack.

>> I I honestly think that might be the future here. Um you guys, I think

future here. Um you guys, I think >> this is not investment advice.

>> No, not investment advice at all. We're

not an RAIA either, so definitely don't take it like that. Um, you guys made a very intentional decision, I think, in the early days when we were talking, which was going after sales teams and

removing the bottleneck to probably the most important thing for a company, >> revenue. Um, how did you guys come to

>> revenue. Um, how did you guys come to that conclusion versus starting off in a number of the other contracts you could have started off with, like employment?

I kind of get teased for my obsession with contracts, but I mean contracts are this like truly incredible creation um that that really resemble code the more you spend time with them, right? There's a lot of if then

right? There's a lot of if then conditional statements that say if this bad thing happens then this part is liable. If this bad thing happens and

liable. If this bad thing happens and that part is liable and to what extent they're really complicated documents and you can see why for technologists, for the engineers on our team, they're a fascinating thing to play with because they're they are they are essentially

there's a taxonomy and there are sense of um rules to them and logic. Why this

matters is the underlying building blocks between all contracts are sort of the same, right? So like if you can figure out payment terms in a three-page um you know quick services agreement,

that payment term paragraph will probably look the same in a 100page credit agreement, right? In terms of like how many days you have to pay us.

And so if you can just figure out all the parameters for that one clause as contracts get more complex, the theory for Crosby is we can do more and more complex documents. So our concept when

complex documents. So our concept when we started was looking at kind of all the types of contracts and we think there's probably about $200 billion spent in the US alone on legal work that are at least contract adjacent reviewing

contracts reviewing ancillary contract documents.

And we wanted to find a type of contract that is really high volume. There's kind

of low risk to messing something up, right? It's not like when you do a

right? It's not like when you do a merger agreement you mess something up, it could be billions of dollars of litigation. But for a commercial

litigation. But for a commercial agreement, it's pretty capped and there's enormous pressure to do them fast. and also just slightly too complex

fast. and also just slightly too complex for AI alone to do them. I think

actually for offer letters or NDAs, AI is getting pretty good. Um but not too complex that AI plus humans um is is you know meaningfully worse off than just humans. And so so these commercial

humans. And so so these commercial agreements, they're about 12 13 pages.

These MSAs were like perfect. There's a

huge demand for speed. Um there's really high volume so you can get great data models on a per client basis. Um, and

they're they're just they're just like they're way too hard for AI alone to do.

When you look at all the AI benchmarks on commercial agreements, it's meaningfully lower than humans alone.

Um, and so humans are still better at them, but humans plus AI, things just get very interesting.

>> How do you feel like your background, having been to law school, having been an attorney, changed your perspective and approach to launching Crosby? You

know, my my parents are both law professors and so I grew up like with this, I think, strong appreciation for the rigor and importance of the

legal profession and the law at large.

What we talk about at Crosby is we want to build a better legal infrastructure.

And a legal infrastructure means that every contract that two people sign creates precedent for the next contract that another company will sign. And

every court case that a judge decides on will create precedent for someone else.

So you have this market of around a million lawyers in the US who are building a public good that every other company and every other private party gets to rely on. And it's just it's a public service. It's really a utility.

public service. It's really a utility.

It's not just lawyers doing things between two parties. It's really for the public good. I mean and when you look

public good. I mean and when you look into the literature on this, it's it's quite outstanding. And so I think I

quite outstanding. And so I think I always had a sense of this growing up like law is important. Law is society important. Law is economically

important. Law is economically important. law is what allows the GDP

important. law is what allows the GDP per capita in the US to be so big because we have a great judicial system.

Um, and I think going to law school just really really really cemented that view.

And so I think what's what definitely gives me pause is I meet founders all the time in the legal space who are who don't have a legal background and are very dismissive of lawyers who think

lawyers are old school, not really sophisticated, don't care about technology. And I think that I have a

technology. And I think that I have a very deep appreciation that lawyers are brilliant, complex thinkers. They care

about their clients. They care about the work they do. And I think that just brings a humility to the work we do that makes uh makes us, I think, more accepted by the clients we work with, especially the legal clients we work with, our lawyers.

>> I mean, it's really interesting to hear that because a couple things come to mind for me.

Your parents are are professors. They're

teaching this profession that ultimately feels like it's in the middle of this like crazy disruption right now. Um, at

the same time, you have a deep appreciation for the work that you're doing. You guys have humans in the loop

doing. You guys have humans in the loop today.

>> I imagine though, long term, the amount of cases that a single human should be able to handle will go from one person doing, I don't know, a 100 cases a week.

You don't need to tell me the real numbers, too. Like, uh, one person

numbers, too. Like, uh, one person handling 100,000 cases a week.

>> Yeah. Um, how do your parents wrap their heads around this and the future of law right now? We actually haven't talked

right now? We actually haven't talked about it too much. I don't think they know what I do, but [laughter] >> but after this, >> oh man, >> I'm doing legal tech.

>> Yeah.

>> Um, but I was actually just on Stanford's campus two weeks ago on the law school meeting with a few former professors and talking about this exact thing and how are they handling legal education. Um, I think Stanford famously

education. Um, I think Stanford famously just launched instances of a couple of big AI like legal AI companies. I think

Harvey gave them an instance and a few others that students can now play with.

So they're taking it really seriously and I think they're understanding that um this is probably a generational change um maybe more than that maybe even a more rare every hundred years the profession changes meaningfully and this

is one of the big ones and I think they're taking it quite seriously. I

think um you know we feel pretty strongly at least this is my point of view that there's going to you know there's sort of a Javon paradox pulling out which is that the access to legal services will become much greater the

cost of every marginal skew of legal work will become much lower and I think you know there's some literature that shows somewhere between 50 to 70% of Americans don't have access to legal

services when they should have right whether that's for things like child support or employment issues you know like personal things or even from a business perspective And so if it gets if a lawyer can now do

10 times the number of cases in a day that they used to be able to I actually think the cost could just go up or at least stay flat because demand will just be exploding and unlocked for for a really great thing that really like vindicates a lot of people that need

access to this.

>> Do we just end up in a world where the bottleneck ends up becoming the government and their ability to process these cases?

>> Yeah, I think this is a fascinating question. I've been thinking for a long

question. I've been thinking for a long time that there should probably be some new type of private um dispute resolution authority that really really is AI native. I know there's a few people are trying to build this. I think

it's a great company idea. I think

courts will figure it out and courts are slow by design and they're expensive by design. You know, I actually spent time

design. You know, I actually spent time in law school working for a court in East Africa for a judge and it wasn't that sophisticated. They didn't look

that sophisticated. They didn't look into every fact. They didn't cite every single case. You could see the rigor of

single case. You could see the rigor of the judicial opinions there was just nothing like it is here. Um, there's

something great about the fact that it takes so many months to get a case through in the US, but you get to a brilliant result. That's just something

brilliant result. That's just something we take for granted, but it's an incredible public good. I mean, truly incredible. I think we'll get more

incredible. I think we'll get more lawyers. I think courts eventually will

lawyers. I think courts eventually will adopt AI. I won't have a choice, but I

adopt AI. I won't have a choice, but I do think there's an interesting place for a private dispute resolution authority like companies already today have arbitration that they go to. And I

think that will probably be a big category. feels obvious I think right

category. feels obvious I think right now because of the fact that you look in other markets like permitting for instance like a lot of states have adopted third-party permit review and

the reason why is because they know that like hey we have a set of rules they're pretty objective here they are you can review it probably faster than us and like we only have so many employees and

I feel like watching that play out which is much further ahead than legal right now yeah we're probably going to be forced to do that here. I think that's right. I think what is unique about

right. I think what is unique about litigation or disputes, there's a very vulnerable sense of trust when you go to court. I mean, two parties fundamentally

court. I mean, two parties fundamentally believe that they're both right and the other one is wrong and one of them is right and one of them is wrong. And then

you you abdicate your right to to to get to some result to a third party that you have to trust. Like you really have to trust and just believe that even if you don't agree with the judge that you know you you consent to their to their

judgment. And there's something very

judgment. And there's something very profound about that, right? In most

countries, you just don't. You disobey

judges all the time or it's corrupt and you can bribe judges. And so that the fact that we have this society is remarkable. And my that's my hesitation

remarkable. And my that's my hesitation with a private entity doing this is that you really have to you're not just selling the right result. You're selling

trust in the process and the person getting to that result. These are hard questions. You don't go to you know 998%

questions. You don't go to you know 998% of cases settle. Those 2% you have to really be at loggerheads to get there.

We could go down a huge rabbit hole when it comes to like biases and like you know uh storytelling in the courtroom and I think what that can do. But to

shift gears back to that trip to India, I'm wondering like what did you learn on that trip? What did you see? I mean you

that trip? What did you see? I mean you realize that there are people without fancy Harvard law degrees contract work for uh the biggest names in American tech, the biggest names in American

industry. What was it on the ground?

industry. What was it on the ground?

What did it look like? You know, I think it was such a lesson in my life of just, you know, it seemed it feel it felt very whimsical to go um to just fly to Bangalore and Hyderabbad, but it was it

was by far the most, you know, um useful thing I did in terms of investigating that one piece of LPOS's.

I think something that really stood out was when I spoke to a lot of people in the US, I think there was a lot of handwaving um and dismissiveness that all these jobs will be replaced by AI in a matter

of months. It's been a year. nothing's

of months. It's been a year. nothing's

really changed. Um, and I think a very probably a somewhat arrogant sense that what they what they do is easy and we can just replace it. And I think what I really found was there was this essential human touch to a lot of the work that's being done offshore. There

are some very formulaic things. There's

truly just looking for a word and keyword searching it and then circling it and you know but you know I remember speaking to one guy and I asked him like how did you get X Y and Z contract closed? and he said, "Oh, I you know, I

closed? and he said, "Oh, I you know, I called I looked up the guy's LinkedIn. I

found his phone number. I called him and he worked for a very big multinational oil and gas company." And he said, and I just said like, "Life is long and your careers are long and we're going to want to work together. You should agree to this term." And I was like, "What were

this term." And I was like, "What were you allowed to do that?" He said, "I don't know. I just I need to get the

don't know. I just I need to get the contract closed. I had a deadline." So,

contract closed. I had a deadline." So,

I called him and I negotiated. Like,

that's so human, right? What a human instinct. And so I think I felt

instinct. And so I think I felt definitely a little humbled by and this is I think led to this idea for Crosby to incorporate our own PLLC law firm to hire Bart attorneys and have the humans

in the loop. Um was just that there is there is an essential um ethereal human touch to legal work that you just can't do without. In other industries you can

do without. In other industries you can in permitting I think you can but in legal there's a human element to it that we just if you do without you're leaving a lot on the table. Does that mean in the future you feel like you guys won't

only be reviewing contracts and legal documents, but you'll also be sending some of that coaching advice that I might get from my attorney at Kulie or Gunderson or Foley today?

>> Yeah. I think someone I was talking to a friend recently who said like when you think about all the things you're a really good lawyer does, right? There's

like advice, there's the word counsel is really fun. They counsel you, right?

really fun. They counsel you, right?

There's sort of a therapy element to it.

They really want you want to feel safe.

You want to feel protected. You want to feel comforted. And a great lawyer does

feel comforted. And a great lawyer does that as well as all the technical legal advice. And so I think if I had to make

advice. And so I think if I had to make some bet on where things are heading in, in a market for legal services, probably you're going to have very leveraged, very expert lawyers. How people become expert is an interesting question I

don't have great answers to, but I think for us it's a fascinating question of how do you free up lawyers to do their most high value work, right? like the

best partners are the ones that you just like I have you know I text my lawyer all the time at [ __ ] and we call all the time just just to like gut check things um that are matters of perspective and intuition

um and I think that's a great place I think lawyers that's a really happy life for lawyers rather than doing a lot of the the drudgery that you know they famously have to do >> so going back to that India though trip where you noticed that there was humans

who were making calls right >> were there any tools that the LPOS's had built out for those teams to help them be more effective and be more accurate I'm sure they built a few things. I will

say what's kind of interesting is, you know, when I got back from India, I was just like totally electrified by this LPO industry. Again, we didn't want to

LPO industry. Again, we didn't want to build one, but we were very fascinated by it. So, I just did a bunch of GLG

by it. So, I just did a bunch of GLG calls with some of the executives in the US of these of these companies. And

>> that was before you guys raised money.

>> Yeah. Yeah. Well, before

>> Wow. coming out of pocket.

>> Yeah. Yeah. Um, thank you to my wife for sponsoring Crosby. um she so we so um

sponsoring Crosby. um she so we so um kept hearing from like the executives about all the AI that was being used and just like how you know don't try to build a company like it's but there wasn't much it was I mean I think part

of the problem when you build a really great um labor and services business that's working is there's not a lot of incentive to innovate and disrupt yourself especially when your margins are pretty thin so you kind of found

this capacity trap so we think about this all the time I think law firms are you know sort of similar going to have to reconcile with the fact that you have a good business you're building by the power and what are your incentives to disrupt yourself and really try to push

a lot of that human work down to automation and computers which which is really hard to do. It's really hard to pick apart if you have a hundredstep workflow and you know step 17 through 30

you can push to automation and steps you know 36 through 41 like having to find the interplay to push to automation is an extremely complex task that I think you probably need like really smart operators to figure out. I mean, the way

you just described it, it makes me think about there's like a the future is going to be a conductor where it's like, you know, you can automate this step to this step and then the work has to get reviewed and then it can get pushed to

the next like 25 automations and then back to somebody else and it's just going to end up driving more efficiency, but humans are still going to have to be in the loop for most things. We we have a lot of like random metaphors we say at

Crosby. Here's one. we say there's like

Crosby. Here's one. we say there's like a paint by numbers element to law and then there's like true you know free form craft beautiful painting and and then there's middling art somewhere you know and right now we can and and but

even finding the things that are paint by numbers and trying to create instructions is extremely complicated and I think most lawyers would just say it's impossible um I think most technologists would say this is all automatable and the truth is somewhere in between probably the most complex

technical puzzle that we've been solving for the last few months is the orchestration problem when a net new piece of work comes into Crosby through Slack Where does it go? How do we break it up? Do we chunk it into sentences,

it up? Do we chunk it into sentences, into paragraphs, into documents? As

context windows get bigger, this becomes a different problem pretty much by the month. And where do we send it? Does

month. And where do we send it? Does

that go to a junior lawyer, a senior lawyer, the partner equivalent, a parallegal LLM or some equivalent? And

in what order? And um giving sort of LMS the context to make those judgment calls is extremely interesting. And I suspect we're just one company dealing with the orchestration problem. But like this is

orchestration problem. But like this is something our best engineers are working on is when documents come in rather than waiting 30 40 minutes to kind of assign it to the right place. Um having an LM do it, you know, within 30 seconds.

What's fascinating about this is a very senior lawyer can look at a request and probably within 30 seconds tell you where it needs to go. But to get an LM all the context from all the edge cases that a lawyer seen throughout their career is actually quite difficult.

>> I want to unpack that more. So your

co-founder who is incredibly talented.

I've met him and uh he's brilliant and was over at RAMP which has done a really good job of >> automating complex workflows. Uh has no background in law.

>> That's right.

>> The good news is you do and I feel like one of the things we've seen in vertical AI is a founder with industry expertise teaming up with uh a technologist where they can like literally sit in the

trenches together and like work on these problems and like figuring out how to solve them. What was your guys process

solve them. What was your guys process like for bringing John up to speed here?

You know, I think it was it was a mutual bringing up to speed by the way, right?

Like I think, you know, so John, you know, is such a unique um type of gifted engineer because he is again going back to the

story of Crosby, he's so empathetic, you know, he is sort of he never thought, oh, lawyers can't do this or like, you know, engineers should do this, but like he really really really and this is what made him, you know, rise through the

ranks so quickly at RAMP. Um was just high empathy with the kind of workflow.

what what what is the what is the user experiencing? When John and I first

experiencing? When John and I first started working together, I mean, he would literally for hours on end sit over stand over my shoulder and just watch review contracts and just I mean, it drove me crazy because I realized I couldn't explain a lot of the things I

was doing. Why did you do that? Like, I

was doing. Why did you do that? Like, I

don't know. It just kind of made sense in my head, right? And it actually made me, I guess, into a better lawyer. Like,

I could really explain myself. Like I

was like, you know, now that you mention it, I saw this thing three years ago and ever since then I've just been like updating this term in that way because it sort of makes sense and nobody pushed back and I think that's the reason there like there's there are very subtle

mental heruristics that lawyers develop like and you're it's just in a millisecond you just apply them and John cuts to the heart of it. He's very

incisive. He's very direct. You, you

know, he's very, very direct, very incisive. And in doing so, he almost, I

incisive. And in doing so, he almost, I guess, reverse engineers the sort of lawyerly process and starts to try to distill out to that orchestration piece what can be automated, what's kind of in between, and what's like fully like we just need a human judgment call on this

one.

>> So, can you walk me through like a case specifically like the contract goes over to you guys? Yeah.

>> You guys chunk it up. Do you send it to one attorney or do you send it to multiple attorneys to like where somebody owns like different sections?

Right. depends on the document. We

usually have right now a lawyer doing the entire document like one just for context purposes. I think you lose too

context purposes. I think you lose too much context when you try to chunk it up, right? Because um contracts are sort

up, right? Because um contracts are sort of hydraulic. If you push on one term,

of hydraulic. If you push on one term, then the other term gets pulled out and if you push on another term, the other so they're they're all connected. So you

can't do them in an unrelated vacuum. So

the context actually matter. We're

starting to realize this more more for MSAs than NDAs. You can probably chunk up a little bit. We usually have contracts reviewed by um it depends on the complexity but by multiple humans um

and so we're trying to just optimize the workflows but I think um having a second opinion on you know there's a lot of judgment calls have to make one judgment call like you know is truly do you

replace an entire paragraph or do you just put in a single word that negates the paragraph right just say not instead of and like these are they're both the same has the same result but you're sort of thinking well what would the counterparty react if they see a bunch

of red on this contract because I just deleted this provision Will that bother them more than me putting in a single word that looks really stubborn and obstinate? So like

these are hard these are like really hard things to teach AI.

>> Are you guys like just to pause for a second? Are you running split tests with

second? Are you running split tests with those now?

>> A little bit but not in a disciplined enough way. We just basically built out

enough way. We just basically built out enough data infrastructure to be able to test things like that right because you can't do you can't do the same change twice on the single contract. You get

one shot. But if you find two contracts that are similar enough then you can start running these experiments. But

even building the infrastructure to track every document that came in, tag it as version five of the an old contract versus mistakenly thinking it's version zero of a new contract um was really really hard. Um but now we do that and so now you can start sort of

getting really creative and trying to test what's working and what's not.

>> That's fascinating.

>> Okay, so you pass it to the the the lawyer. There's two people that review

lawyer. There's two people that review this thing, right? Are they like is the LLM like running the first review for them and giving them like the track changes that they're accepting or

declining? Is there are there insights

declining? Is there are there insights on the right hand side of you know it probably originally I'm starting to get more complicated but it looked a lot like any other Word add-on you can see on the market right like that is just on to lawyers that they

download Word and it kind of suggests changes for them. I think what LLM are great at is recall so they catch all the changes that a human scheming a document might miss. They're actually over

might miss. They're actually over inclusive. they might just like have

inclusive. they might just like have some sort of false positives. They might

just say like this is a thing you should look at and you're like actually that's fine. I don't have to review that. Um

fine. I don't have to review that. Um

less good at precision. By that I mean they'll they'll they'll catch an issue but they won't give you the right result. Or they'll like have a bunch of

result. Or they'll like have a bunch of words for this for this change that you again you just need one word or two words. And so they flag all these um

words. And so they flag all these um issues to lawyers. And so rather than having to read every word on a page, you can just jump from one issue to the next. And that's like you you can

next. And that's like you you can imagine how much time that saves, right?

You might jump across five pages and just not have to read those. Um, so

that's what the elements are great at.

Now, I think the big unlock has been building very comprehensive knowledge bases for each of our clients. So, when

you send something to a lawyer, they send it back to you with a bunch of comments with your name, Jason in the margin saying, "Jason, what do you think about that? You let me know what you

about that? You let me know what you think. Go comment on it." If you tell

think. Go comment on it." If you tell your lawyer like, "Listen, our subprocesses are all in the US. They're

not abroad. You can answer that question, but you know, if you see it again, they'll typically forget in a few weeks." And so, we commit that to a

weeks." And so, we commit that to a knowledge base. So, that's like the big

knowledge base. So, that's like the big unlock. And this there's like literally

unlock. And this there's like literally like thousands of examples where we ask a client, hey, how would you answer this question? Can you comply with Hungarian

question? Can you comply with Hungarian law? We've never heard that before. And

law? We've never heard that before. And

they say no. And we just remember we never have to ask you again. And so this AI toolbar that we're building is bringing all these really really discreet um learnings to the review and helping the lawyers work faster. I think

the other piece is building a comprehensive um market benchmark um sort of database.

So, because we see so many different clients and so many different counterparties, we really know what like, you know, a mid-market SAS company in a really large Fortune 100 will agree to from time zero. And we just aggregate

those, but we can just tell you guys like just agree to these payment terms like you're you're going to spend turns, just disagree, just agree to 60 like you'll get it done today. And and so that's something that we've been getting more and more invested in. And again,

that pops up in a word out on as a lawyer sees it, sees the citation, where it came from, why we made the decision, and decides, okay, I'm going to execute this change.

>> Yeah. It's like us uh storing all the red lines of term sheets that our portfolio companies have had against other venture firms and figuring out where where might we be able to push.

>> Yeah. You don't want to red line against primary. Uh you know it's all for the

primary. Uh you know it's all for the founder benefit at the end of the day.

>> Exactly.

>> Um okay. So you guys are going through this process and you mentioned something earlier that I just want to like peel back a little bit more which is there is context that's like obvious you know

that could be stored and already has language around it and can be found in like a generalized LLM but then there's a lot of context in people's brains and I was talking to uh an executive over at

Distill and we were having this conversation because all these companies have forward deployed engineers now and they're historically they've gone in they've sat down next to the people exactly like John did with you and

they've tried to not only take the context out of the systems but they've tried to take the context out of people's brains.

>> Now they've developed technology to help with essentially like trying to capture more of that through the computer you know screenshots and so on and so forth.

>> How are you guys like pulling that context out of the expert's brain?

>> I think that is that is the crux of AI application companies. That's it. How do

application companies. That's it. How do

you extract things out of a subject matter expert's brain and hand it off to um the computer? We have a lot of ways we try to do this. I think like a really simple unlock that we did was we just

changed our seating arrangement a couple months ago to have lawyers and engineers, lawyers and engineers um offset one by one by one. Seems trivial,

but there's a lot of looking over at the person's screen next to you and seeing what's going on. And it's actually helped a little bit um just for like understanding the process. I think there is a lot that you really can't get out

of someone's brain. It just there's just there's just extraordinarily subtle judgment calls that lawyers make and that's why we just are trying to build a system that includes it in the loop.

Now, can we take them from an hour to 60 seconds for each, you know, piece of work they do? I think that's quite reasonable. But can we remove them all

reasonable. But can we remove them all together? I don't think that's even the

together? I don't think that's even the point. I think you're missing the point

point. I think you're missing the point there. And it's a bit of a fool's er at

there. And it's a bit of a fool's er at least which at least we're the state of the artist today. Part of the key of what we're trying to do at Crosby and why we're able to do these really fast turnarounds. We're trying to make the

turnarounds. We're trying to make the lawyer reviewing fungeible. So, at most companies, you have one lawyer who's assigned to you and if they're busy, you're not going to hear from them for three days, but they remember some things about you. They forget. They

forget, but they remember some key points and they have their own mental models that they rely on. And so in our mind, can we show just enough information to sort of jog your memory such that like you know each client can have a handful of lawyers that are

assigned to them and whoever's available and will be available because we can model out sort of availability and demand and supply um within a few seconds. So that's I think the key is we

seconds. So that's I think the key is we need some human ability to come in and make judgment calls and rely on their own heristics but we jog the memory enough by trying to extract what we can

but I think that you know um human minds are are are much more powerful than LLMs just like right just is what it is.

>> Yeah. Yeah, I mean I think there's uh there's two levers there, two things there I I want to chat about, which is one, and you don't need to get into numbers, but I'm just wondering like how

do you measure accuracy then for your teams and benchmark that to make sure that you guys are improving over time?

>> Yeah, so we measure really uh three things. We measure turnaround time um

things. We measure turnaround time um which is the time a contract comes into our inbox to the time it leaves our outbox. And we add that up between the

outbox. And we add that up between the first review and the second review and the third review. So we call it ttat, total turnaround time. And it's

important because the best way to reduce tat is just doing fewer turns, right? So

if we can give a really, really, really smart review and then we only have to do two back and forth because the counterpart just accepts it and we sign, that's the best way to reduce the third, fourth. Okay, the second one is hurt

fourth. Okay, the second one is hurt human review time. And we literally just look at sort of in aggregate how much time from the time a lawyer clicks open a document to then they click close. We

track all of it. How much are we reducing that month over month? And um

you know t can be again it can be impacted by doing few reviews. It can be impacted by just like reducing Q times weight times for a human. And so that's an operational question and also by by reducing the hurt. So that's just time

quality. Um in legal work it's this like

quality. Um in legal work it's this like really beguiling issue because actually when you look at the academic literature there is some line of thinking that says it's

impossible to measure quality in legal work and [snorts] this goes back like 30 40 years. In other words, if you ask

40 years. In other words, if you ask somebody, why do you pay that lawyer $2,000 an hour instead of that lawyer $1,000 an hour? Is he really two times better? It's probably impossible to

better? It's probably impossible to explain why. And it's probably

explain why. And it's probably tautological. You probably say, I pay

tautological. You probably say, I pay him more because he's really good and he's really good because he charges more and it's sort of circular. And so what we've been trying to do like obsessively, and it's it's a really it's a fascinating computer science problem

is distill out what is quality, what makes good legal work. And 10 lawyers, the more senior they get, actually have less agreement on what quality looks like. they start to break out into

like. they start to break out into slightly more um opinionated versions.

>> That's pretty fascinating. And I think what's interesting is you guys are amongst this like larger crop of human in the loop services. You know, we help

launch a company called Table that does it for QAQC construction drawings. We

have another portfolio company called Valerie Health that does in healthcare.

>> Yeah.

>> And I feel like a consistent conversation that's happening in every board and you know, we talk about it as partners is like, >> okay, how should we think about cost to

serve and where are we willing to run things at right now? If you were giving advice to

right now? If you were giving advice to a founder today that had a human in the loop type of company that was optimizing for what you're talking about, um, how should they think about cost to serve

and whether or not they should be running at negative 20% margins or 10% margins or 90% margins? My first piece of advice is talk to you. Talk to Jason.

Talk to Jason at primary if you're starting a team in the loop business. I

think like from the get from the jump you guys really understood this space and so it's always been such a joy to talk to you about it. Sincerely, I think there's a lot of investors who I think reasonably say well this isn't quite software and so what are we doing here?

But we had this we had this strange moment where for huge swaths of human labor there's finally technology that can um augment and replace a lot of that work. It's just it's kind of

work. It's just it's kind of mind-blowing. You know the margin

mind-blowing. You know the margin question is I think very company specific and very use case specific. I

think there are types of human labor that you can replace today but the amount that your clients will pay you may not be more than the inference costs of serving those clients right just like that's just like raw raw raw math.

There's another strategy that we contemplated in the early days. We

decided against it which was let's really char let's like compete on price and just hope that you know um LM will pick up enough that we can remove humans enough that we can start to make a

margin here right and just like be margin negative and just undercut every law firm um with the hopes that even though AI is not there yet today we'll figure it out in a year or two which is I think a pretty aggressive and ambitious and interesting company idea.

I think for us the idea of competing on price in in law just starts to um be synonymous with low quality and probably means we have to hire worse lawyers and we just weren't willing to compromise.

Just that's specific to law though. I

think law you know unfortunately you get what you pay for and and good lawyers are more expensive and they understand that about themselves and we wanted to hire those people.

>> I think what we competed on was speed and our speed is like I mean it's unmatched, right? like a two-hour

unmatched, right? like a two-hour contract turnaround no matter what is is pretty crazy. And we've just we've just

pretty crazy. And we've just we've just held that firm no matter, you know, as we scaled. And we have a great bet that

we scaled. And we have a great bet that our margins will go up, but we always wanted to know that we were charging enough that we were like, you know, charging kind of what our value was to our clients and probably undercharging, right? We could probably say, "Hey,

right? We could probably say, "Hey, we're going to charge double your law firm because we're so fast and we have market insights and we remember things about you." Um, so I'm sure we're

about you." Um, so I'm sure we're leaving money on the table, but that's that's for another day. Um, I suspect it's just how how, you know, so I think it's it's somewhat industry dependent and somewhat the psychology of the customer, right? like do they expect to

customer, right? like do they expect to pay more for much better um human input?

>> It it is a hard decision to make, right?

If you're making a bet that your margins are going to improve dramatically because the LMS are going to get better.

>> Yeah.

>> Uh >> that can be a hard bet. It's not in your control, right? And so from a core

control, right? And so from a core competency perspective, I feel like it's really important to analyze like what is in your control and like what's not in your control. if you know where like the

your control. if you know where like the context is coming from and you know the exact steps to get there and sure there's technical risk but there's not

like anthropic risk I think I feel a little bit better about that but what's also fascinating is at the same time the venture markets are rewarding growth and so I always am telling founders now I'm

like if we feel really confident that we can figure out the context challenges and the technical challenges I don't care that much about margins right now.

Like let's just prove there's zero demand risk and we need to be thoughtful about utilization from like an employee staffing perspective because you don't want >> three attorneys sitting in your office

in Soho, you know, hanging out and twiddling their thumbs. But like as long as you're delivering a high quality service and you show that there's demand and there's a pre-existing market, I'll bet that the margins will go up over

time. But I need to be able to see that

time. But I need to be able to see that clear path. I often talk about this sort

clear path. I often talk about this sort of execution risk and market risk and like SpaceX was pure execution risk. If

you can build cheap usable rockets like there's an enormous market for that, right? But can you do that? Whereas like

right? But can you do that? Whereas like

Snapchat, you know, has like execution risk is low. I mean, it's ephemeral messaging. There's some scaling issues,

messaging. There's some scaling issues, but basically it's you can it's not that hard to build, but the market risk is high. Like do people really want

high. Like do people really want ephemeral messaging? And that's a huge

ephemeral messaging? And that's a huge question mark. And then sort of I think

question mark. And then sort of I think I try to think about every company as somewhere falling in between that spectrum. I think services are mostly

spectrum. I think services are mostly about execution risk. Do people want to pay lawyers for legal services?

Obviously like look at the market and we're not trying to create something altogether different yet although we're working towards it. We're trying to create something better. So there's a market there's demand like it's just there is demand and I think we

oftentimes get asked by investors and candidates like will people really pay for this? And we're like just like look

for this? And we're like just like look there's a trillion dollar spend of lawyers. We're just off we're just

lawyers. We're just off we're just selling it but better. Of course there is. The question is execution and so

is. The question is execution and so actually I think the margin story is more interesting for a lot of these companies right I think the the the topline growth is relevant because you have to understand does the branding

work does the distribution work are people willing to pay Crosby dollars that have they've been going to a law firm that you know has all the lawyers on their website so okay so you can derisk that just like will they will they trust your brand and then the only

reason I think growth sort of numbers matter is probably there are some industries where distribution matters more than anything else even today right like you're just ability to be the

contract council for all of these companies as the models catch up is probably important. I think Harvey's

probably important. I think Harvey's done a phenomenal job, truly fantastic, of just infiltrating all the law firms, the big law firms and being the trusted brand for LLMs. And I think in the early years folks there I've talked to all the

time admit like it was really distribution story and then now the models have caught up. So it just depends on the market and kind of if is there like a true distribution advantage that's durable.

>> Let's talk big law for a second.

>> Let's talk big law. Uh man, I feel like it's obvious in some ways and you hinted at it, which is like the future of big law is probably higher leverage work by the attorneys.

It likely is probably more projectbased work, right? But if we look back, the

work, right? But if we look back, the last generation of like legal tech companies, I wouldn't call them the massive successes. Um what's changed

massive successes. Um what's changed today? Why are law firms now looking at

today? Why are law firms now looking at buying products that might literally reduce their billable hours?

There's a great article that just came out um from a Georgian professor a couple of weeks ago talking about how um he makes the argument that bill hours will get more expensive because if you

can do more work in a single hour then you should be charging more for it, right? Because I'm only going to charge

right? Because I'm only going to charge three minutes when I used to be charging 20 minutes. I think that's that's

20 minutes. I think that's that's probably reasonable. I think big law is

probably reasonable. I think big law is not going anywhere. Again, those top hundred law firms have an enormous market share um and really do the most sophisticated lucrative work. And part

of that's branding. A lot of it, you know, law is a credence good, which means you can't measure the quality of the legal service until you've consumed it. And only if you're an expert, right?

it. And only if you're an expert, right?

You need another lawyer to look at some piece of legal work and say if it's good or bad. And that's really crazy, right?

or bad. And that's really crazy, right?

Like think about all the things you buy on a given basis. And you you typically know what you're buying before you pay for it. So you have all these

for it. So you have all these intangibles to try to give you proxies for is this good or bad. And big law firms have all the best intangible assets. They have brand, they have

assets. They have brand, they have credibility, they have a track record.

So I just think that's not going anywhere. But again, I don't want to

anywhere. But again, I don't want to just say AI, but I mean just software and computers couldn't do anything relevant for law until three years ago.

I mean like they could help you store things better. And I think the most

things better. And I think the most significant innovation for law firms, I say this ironically, was Microsoft Word, right, almost 45 years ago. Um that

really changed things like I had lawyers and you know there's like small things.

Docuign helped, Carta helped, um, IM manage helped, online research for for Breeze helped, but these are all like these are, you know, kind of marginal things just chipping away on the edges.

And in all these cases, they probably kind of explain why the billable hour has gone up because lawyers can be slightly more effective. But now we're talking about like the meat of what legal work is of just drafting, reviewing summarizing researching and

revising being completely offloaded to computers. That's like really dramatic,

computers. That's like really dramatic, and I think law firms are going to have to figure that out. I will say the partnership structure of law firms does not allow for capital expenditures

in risky innovation. What that means is when you're a Ccorporation, like any startup you invest in, you can sell a percentage of the company and try some bet and if it doesn't work out, then there's no recourse. For law firms,

either partners say, I'm going to take home less money this year to allocate towards capex, or I can take out a personal loan recourse against me, or not personal, but a loan recourse against me um to invest in capex. And so

it's really really hard and you have a vote of the partners and the older partners who have more authority are the ones least likely to see return on this investment. So it's really hard for them

investment. So it's really hard for them to allocate money that they could take home today um to long-term innovation.

So that's I think the bigger thing than billable hours that people aren't sort of aware of. So if you have this like double incentive structure challenge which is one the partners need to vote

on this capex thing and two our model is going to be changing and who the heck knows it might take three or four or five years for that paper to be proven out but in the meantime maybe your revenue does shrink a little bit.

How has it been possible for the companies like Harvey and Lora and the others to just grow so quickly? At least

in 2024 was every law firm had to had to had to be trying something and they were right to you know you can't not be trying different AI solutions when they're so meaningful. I think this year and next year we'll show that lawyers

are actually using it and they are I mean I talk to lawyers all the time especially the junior associates they get real value from it. The reason why we incorporated our own law firm and have our lawyers who are using our product sitting next to our engineers is

because we don't want AI to just be doing 10% of their workflow. We want to offload as much as we want to offload up to 80% of the workflow. And this is rigor. This is hard. This is like

rigor. This is hard. This is like painstaking work that most lawyers we interview just can't figure out. And we

they don't pass our process. Used to be very thoughtful, cerebral, self-reflective, kind of be able to figure out like, oh, you know what?

There's that thing that I've been trained for eight years that only I can do, but actually I can offload it. um

it's just not what lawyers are good at.

And so I think there's going to be a real I think there's going to be a place where companies like Crosby and and and other um you know legal service companies um really explode because we're able to capture and figure out innovations in what you can offload to

computers that law firms can't. I think

they'll still command the most uh um coveted parts of the market, the most high value M&A, high stakes litigation.

You don't really want to offload that to automation anyways. That's like just

automation anyways. That's like just you're not sensitive to price or time.

You just want the best. But for

everything else, there's going to be enormous new legal companies that have better ways of doing things.

>> I feel like I've heard arguments where it's like, okay, big law will be around.

They're going to focus on the high lever things. Um, you guys are approaching it

things. Um, you guys are approaching it where you're essentially a net new outside council in a way. And then

there's other companies who are going and selling software directly to the GCs in these companies and saying like we're probably going to lower your third party

legal bills and everyone's cutting the market up in different ways. What do you think people are getting wrong about this whole let's go sell to the in-house

GC which by the way like Harvey's also going after in-house GCs now too. So I

I'm just wondering what is the future of in-house GCs and what do you think the companies that are trying to sell directly to them their own software are getting wrong?

>> I actually think they're getting it right. I think in-house makes the the

right. I think in-house makes the the ascensions are much more aligned. I

think the secular trend over the last 15 years where in-house teams have gotten much bigger. You know in-house teams

much bigger. You know in-house teams went from 340,000 to 430,000 people um from 2014 to 2024.

um from 2007 to 2017 I just these numbers are [snorts] interesting to me they grew in-house teams increased 200%.

Whereas law firms increased 30%. Right?

So you have like these enormous in-house teams now. Google has a thousand

teams now. Google has a thousand lawyers. That's bigger than I think

lawyers. That's bigger than I think everything other than the top 50 law firms. They have a huge law firm within Google. Now the reason why this is

Google. Now the reason why this is interesting is law firms that are selling legal services to a bunch of clients have a lot of restrictions on how they can organize themselves. you

have to be barred attorneys in order to work their own equity in the company, right?

>> Um, did that change recently by the way in some states, but it's not really it's not it hasn't been catching up as much as we thought it would.

>> Uh, you can't get and so and so like nonbart attorneys can't own equity and own upside. And then there's the capback

own upside. And then there's the capback issue I talked about a second ago.

Whereas for an in-house team, if you have one client, so Google's thousand person law firm, um you can have all sorts of people in this entity that are doing legal work or legal adjacent work and you're kind of uncapped by the

regulatory issues because you have one client. And this is an incredible

client. And this is an incredible anomaly. And so what you find is when

anomaly. And so what you find is when you look at a corporate legal team, only about 60% of the people there are lawyers and you have about 20% are parillegals and 10% are op staff and kind of the difference of these like various roles that you just wouldn't

find in a law firm. In other words, like when you look at Google's legal team, um it looks very different than a law firm.

And I think what you're seeing now is a convergence on so you're bringing more and more work in house. So I think selling software to in-house teams makes a lot of sense. Um in some ways what we're trying to do is create something that feels like a service, but fundamentally what you're getting here

is mostly software from Crosby. And I

think that's the right I think that's the right trend. And I think that means law firms are more specialized. They're

more expert. Um you really have to come for them for things that you cannot hire for internally or it's just too expensive. It also just feels like

expensive. It also just feels like there's a really good market segmentation exercise to be done which is like Crosby went after the cursors of the world and like the really early days

and as that company scaled they were already using you. Yes. And your net revenue retention probably looks incredible on a graph with all these customers because you know similar to

Shopify who makes a lot of their money if not most of their money off of the 10,000 20,000 Shopify plus customers that are growing. They grow with them.

you guys are going to experience the same dynamic here and maybe they never have to bring in a huge in-house team versus if you're going and selling to the incumbents who have a thousand

people on their legal team. Maybe if

you're a founder today, you want to go in, you want to look at what does each one of those people do because I think we have a mutual friend who used to be the GC at Row. Yep. And I remember he told me he had like multiple people on

his team that were just reviewing marketing and advertising materials.

>> Yeah. because in the health care space that's a really important job. You don't

want to get sued or shut down by the government. And it feels like there are

government. And it feels like there are just like little things that you can kind of pick off in the same way that we're kind of seeing it done with big law where the Harveys and Loro started

like horizontally I think about and then you've had companies like Wexler that are going really deep into litigation or manifest law which is going really deep into the direct to consumer

interestingly but mainly on like immigration side. And so there's like

immigration side. And so there's like all these ways to cut it up to your point, but I think that people can't just look at it by like law type, but actual customer size too feels critical.

>> Yeah, I think that's totally right. You

know, we have one of our very early clients, I think they were four or five sales people when we first started working with them and now they're probably 50 plus. And they have a couple of internal lawyer, they have two internal lawyers now. And one of their

lawyers said without Crosby, we'd be 20 lawyers today, which like the ratio is um I mean I think sorry, they're like over 80 sales people now. the ratio is like a little high, but like that sounds right. And they have this like they sort

right. And they have this like they sort of took this um you know inflexible labor cost that's just like cost of lawyers and we're able to turn into more flexible cost with Crosby. They just pay

us for the contract. That's amazing. And

it's also really fast, right? An inhouse

team that's not an enormous company that's basically that's anything other than the top hundred biggest companies will not be able to invest in innovating on legal as a function. They will not

figure out the best algorithms to send documents to what person. That should

not be their center of excellence.

>> So you have companies like Crosby that we're going to be the best place to handle these contracts super super fast and creatively. We'll do it for you. I

and creatively. We'll do it for you. I

think our positioning has been very deliberate from the start. We are the AI law firm for companies that move really fast. That's it. It's not for a certain

fast. That's it. It's not for a certain type of company. And what's fascinating is a lot of very large corporates have been reaching out to us in the last few weeks saying like we move fast. Like we

we genuinely have like a very different function here internally. We see that you're working with these great, you know, SAS companies. We want to try you out. And and they and and they mean it,

out. And and they and and they mean it, right? And so they're taking a pretty

right? And so they're taking a pretty rigorous approach to reassessing how big their legal team needs to be. The kind

of irony is that legal teams stay the same. They can just move a lot faster.

same. They can just move a lot faster.

They never they never reduce headcount, right? I think it it is a little scary

right? I think it it is a little scary at first, but you know, most legal teams are too small at every company we've seen. they're backlogged, they're too

seen. they're backlogged, they're too busy. And so we think of ourselves as a

busy. And so we think of ourselves as a second set of hands. Um that just unlocks to move a lot faster. And the

NPS of these big legal teams when they use Crosby goes up. Like all the stakeholders across the company, whether that's procurement side for buying software or the sales side for selling it are really delighted when their lawyers turn things around in a few hours because they're using Crosby.

>> I feel like it's not surprising. I

remember when I was a chief of staff at a software business, the biggest complaint that the sales team would come and like grab my shirt about was like, dude, like why haven't we turned around this contract yet? And like they feel like it's going to hurt them because

they're trying to hit a quota and they're trying to get paid. And like if you can do it faster and less turns, then it's going to make it a magical experience for everyone involved, >> right?

>> Um, so I want to talk a little bit because we don't have that much more time uh about your hiring process. I

think you and I we had dinner I don't know a few months ago uh on on your rooftop we were grilling and we we kept talking about hiring and then interestingly uh you canled this first interview uh because you had to go focus

on a hiring related thing which by the way >> I love the obsession with hiring. I

think like the best people like the best founders I work with >> are just maniacal about it. Yeah.

>> And John came over to work with you from RAMP. I feel like culturally they have

RAMP. I feel like culturally they have an incredible hiring process. What have

you learned between, you know, being a lawyer, then running, you know, and helping run another startup to now?

Like, how has your hiring process changed? I'm really proud of the hiring

changed? I'm really proud of the hiring we've done at Crosby. I think RAMP is unambiguously the best talent dense company in New York City on the engineering side and John was, you know,

first 15 engineers there and brings a lot of those learnings. It took us about five months to hire our first engineer.

And it was it was it was it was tough because we had so much we wanted to build, but John was unrelenting in not wanting to reduce his bar because that first engineer governs the next 10 15 engineers that you hire,

right? The quality you have there. And

right? The quality you have there. And

that was the right call. Now we have um many more. I think being in a market

many more. I think being in a market where there's a lot of venture dollars being poured in means that labor market is pretty tight for great engineers.

It's really competitive. But it also forces you as a founder to sharpen your instincts and become excellent and maniacal at finding the best people. And

um it's probably the most important game to play if you're trying to build a very serious company today is find ways to hire the best people and you should be competing with only two or three great companies. And that's kind of where we

companies. And that's kind of where we found ourselves. I'm very proud of it. I

found ourselves. I'm very proud of it. I

think we probably have top two or three talent dense companies in New York City and that's amazing. I I did not think we'd be here a year ago and we we really are. One is

are. One is we really honestly had to figure out what is our edge and that was um finding people who are former founders or

aspiring founders and aligning with their ambition to be a founder again or to be a founder in the future and convincing them seriously that this is the best place that they're going to be on that path. So our first

five founding engineers were all former founders. One of them went through YC

founders. One of them went through YC and Race Van Andre. Another one was starting YC. met him like they were all

starting YC. met him like they were all like really starting companies and they just felt like they didn't have enough conviction on an idea and they're different cases but for one of them at least didn't have enough conviction the idea wanted one more rep wanted to learn

from John and me and um it's it's it's a privilege to be able to work with these guys and it means that you have to you have to own you have to own up to that promise that means bringing them to customer meetings when you get invited to a founder dinner send them in

yourstead it means exposing them to different investors so they can kind of understand how that works it means giving them a lot of ownership on products so that if those fail, they're going to be publicly accountable on LinkedIn. We will we will call them out,

LinkedIn. We will we will call them out, but if it succeeds, they get all the glory because that's what being a founder is. You're building them public.

founder is. You're building them public.

It's really stressful and I want them to feel that. I think because we've done a

feel that. I think because we've done a good job in the first year of building a reputation for incredible people that we have no business hiring because they're former founders or aspiring founders. Um

we've built a really good machine for bringing great people in.

>> What does an interview look like then for you?

>> You know, for an engineer, there's a typical tech test stuff like it's all kind of standard. Um, but we do pretty intensive personality deep dives to understand like truly what are your goals.

When you're hiring people who have ambitions to start something one day, they're pretty missionoriented. It's not

that financial in some ways. Like I, you know, I suspect they'll all do well with Crosby, but this is just a stepping stone to whatever the big win is that they're working towards. And I'm totally agree with that, right? But I I I want them to feel the, you know, urgency and

intensity of a great company. And so we owe it to them. Like I I mean I feel a lot of weight on my shoulders that we owe to our team, you know, because they took a chance on this early company and they're building something great. But in

the legal domain, I think there's never been truly exceptional, you know, best in the world engineering talent trying to solve these problems

until right now. And this is going to be the differentiator. Like I like I like I

the differentiator. Like I like I like I no longer really spend that much of my mental time thinking through all the innovations. It's just about how do we

innovations. It's just about how do we unlock the people in our team to solve those problems? And that's a really big

those problems? And that's a really big shift as a founder.

>> Are there any go-to questions you've liked to ask in your interviews? Now,

>> I ask every well, this is my secret sauce. I I personally ask every person I

sauce. I I personally ask every person I interview what's a book they've read recently that changed their thinking and now they're all going to have a prepared answer, but it's a great question and some people are very cerebral and it's always surprising.

>> I I love that question. I I tend to ask people too like what's a pivotal moment in their life or pivotal person and like walk me through how that changed the way you look at things and and you learn a

lot through those questions I find. Um

so that's the engineering side and then on the non-engineering side do you have a different mental model or is it a very similar process?

>> Um so the other two pillars of the company are the ops team and the legal team. I'll talk about both. We

team. I'll talk about both. We

give offers to an extremely small percent of the lawyers that we meet and we get, especially as we've made more noise in the community, a ton of inbound from amazing, you know, top five law schools working at a top 10 law firm in

New York. This is a great place to hire

New York. This is a great place to hire lawyers. It's in New York City. And a

lawyers. It's in New York City. And a

lot of them, you know, fourth, fifth year attorneys looking to leave big law to go in-house. It's kind of a typical turning point in a career. Are not that missionoriented and it just isn't going

to work. like they like legal work, they

to work. like they like legal work, they enjoy the profession. The ones who have gotten who have joined us, they see the software building and it's like a light switch turned on. Like they're just they're just they're electrified by this

idea that they can build a better legal infrastructure. That's it. Period. Like

infrastructure. That's it. Period. Like

that gets them going. They work their tails off. They're like they were the

tails off. They're like they were the hardest workers in the company. The

lawyers that are at Crosby are working late into the night. They're building

data taxonomies. They're doing product work. And they're looking at a ton of

work. And they're looking at a ton of contracts and doing quality work. And

finding people that are at mission oriented is extremely rare and it just means you can you can you you sacri you sacrifice time to find those people candidly, right? If we hired more people

candidly, right? If we hired more people more quickly, we wouldn't have that bar and that's okay. I think it's worthwhile. And then we have this ops

worthwhile. And then we have this ops team which has quickly become like this total gem in the company. A few of them came from scale. Um we hired over the summer. Um they all have technical

summer. Um they all have technical backgrounds. They have CS degrees.

backgrounds. They have CS degrees.

They're not engineers, but they know how to code. And in the last few weeks,

to code. And in the last few weeks, we've noticed that they just keep going and seeing that lawyers are doing things that are just taking six clicks and they should take one and they just build like an NADN or Zapier or like really quick fix for it. And we have like dozens of

these now and they're compiling and they're really starting to compound. Um,

that's been a really special team to see in action is kind of just like this hit squad that just bandies about without much of a road map every few weeks and ships a bunch of quick wins that really stack up. I think that's probably

stack up. I think that's probably something I wish I'd known a year ago was for tech enabled services businesses having like a cracked hit squad that's technical but really operationally minded changes everything.

>> Yeah, it's so so important. We've seen

that a lot at D&D for sure. Yeah.

>> All right, last three rapid fire questions for you. How have you changed the most as a founder?

>> I think I'm trying to learn how to become more hands-off and trust the people around me are for sure smarter than me. I mean truly are. And my job is

than me. I mean truly are. And my job is just getting the great people in and keeping them inspired and not trying to do things myself. That's been really hard >> in your back channels. That's what I

heard too, by the way. So, [laughter]

uh, number two, you just got married.

>> Yes.

>> God blessing. You know, hopefully you'll have a kid one day. Um,

would you advise your kid go to law school?

>> I think I probably would. I think law school was one of my favorite three years of my life.

It was so intellectually interesting and I genuinely I always say this but it really is the golden age for legal innovation. What lawyers are going to be

innovation. What lawyers are going to be able to do the leverage like their importance in society is going to be even greater than it is today. I think like we're just we're

today. I think like we're just we're just at the beginning. Like this is so exciting.

>> And last one I ask every guest this question. If you're mayor of New York,

question. If you're mayor of New York, what's one thing you would change?

>> I think we should have pizza on every block. Like pizza is 60% of the reason I

block. Like pizza is 60% of the reason I live in New York City. It's best in the world. But there's not enough more

world. But there's not enough more variety. You should always be like

variety. You should always be like walking distance from a pizza shop.

>> Your co-founder would love that answer because I'm pretty sure when you guys started the company, what you guys had pizza like every day for like >> Oh my god, 25 days straight. That's

that's real founder mode right there.

Well, I appreciate you joining me, Ryan.

Thanks for coming in and uh excited to see what this becomes.

>> Thanks, Jason.

>> Appreciate you carving out a slice of your day to think a little bigger with us. Do me a quick favor. Hit follow,

us. Do me a quick favor. Hit follow,

drop a fivestar review, and share the episode with one friend who's eagerly obsessed with building future. Huge

thank you to the team that makes every release sound effortless. Atomic Growth

for the behind-the-scenes horsepower, the brilliant Katherine Theer for steering our editorial compass, and Devon Naton for getting each episode into your feed right on time. Their work

lets me focus on bringing you candid stories and gritty tactics from the founders, operators, and investors, rewriting entire industries. Until next

time, keep challenging the status quo and I'll see you on the next change order.