Learned Hand’s Shlomo Klapper on Why Courts Are the Next Frontier for Legal AI

On today's Law Next, I'm joined by Shlomo Klapper, CEO of Learned Hand, an AI company building what he calls a reasoning engine for courts.

Shlomo's path to founding Learned Hand winds through the law firm Quinn Emmanuel, where, as an associate, he became an early power user of case text.

And then Palantir, whose approach to category creation clearly informs his current venture.

Now he's tackling one of the most consequential challenges in legal technology, helping judges manage overwhelming caseloads at a time when AI is making it easier than ever for litigants to file cases.

This interview comes on the heels of significant news.

Learned Hand just announced a partnership with the Superior Court of Los Angeles County, the largest trial court in the nation, to explore how AI can support judicial officers across the full arc of a case from filing through drafting.

In our conversation today, Shlomo and I discuss why courts are the next frontier for legal AI, what it takes to earn the trust of judges, and how Jevons Paradox, the idea that as legal services get cheaper, demand will explode, is reshaping the justice system.

I am Bobby Brogi and this is Law Next, the podcast that features the innovators and entrepreneurs who are driving what's next in law.

Before we get to any of that, please take a moment to hear from the sponsors who so generously keep this podcast running.

Managing an immigration practice across multiple languages shouldn't mean double the work.

Lollilaw's comprehensive platform offers clients, questionnaires, and payment portals in over 20 languages, instant form updates, and automated payment processing.

Create invoices in seconds, track payments effortlessly, and streamline your entire practice with the number one rated immigration case management software.

Experience the difference at lolilaw.com.

That's L-O-L-L-Y-L-A-W dot com.

You didn't go to law school to waste time writing and responding to discovery requests?

Never waste time on discovery again with BriefPoint.

BriefPoint drafts your discovery responses and requests for documents within minutes of your first sign-in.

BriefPoint is the most reliable discovery drafting solution on the market.

and it works right out of the box.

Get your first document drafted by going to briefpoint.ai and clicking get started.

No credit card required.

BriefPoint, draft what matters, automate the rest.

Now let's get to today's conversation.

Shlomo, welcome to LawNext.

Bob, thank you for having me.

I'm really looking forward to talking to you.

I do have to kind of quickly share the story, only because I think it's so funny that just a few weeks ago, I had Pablo Arredondo on this podcast.

We were sitting out in San Francisco chatting about things.

during the course of that recording, as well as offline, he said, he dropped your name a few times saying, Shlomo Klapper is somebody who's really doing some cool stuff and you ought to talk to him.

We had never met.

And then last week, I'm kind of wandering toward my hotel while I'm in Manhattan for legal week and I hear somebody say, Bob Ambrogi.

And uh there you were standing on the sidewalk as if fate had delivered us or something.

uh And one thing led to another and here we are chatting today.

Well, I'm glad that fate did deliver us and Pablo is my hero.

I would not be in legal tech if it were not for Pablo.

So I'm honored uh beyond words that he's the kind of thing.

it starts with when I was an associate at Quinn Emmanuel.

I had discovered this tool called Case Text and I initially discovered it because it had this little nifty thing where you can copy it and it would give Google citations.

And then I started looking at some of the other features that they had and they had this wonderful thing called parallel search, which was this, this was in 2021, 2022.

And it, Bobby gave me superpowers.

I had all these partners over the firm emailing me.

like, what is this that you did this?

Find me a case that stands for this.

I literally just take it and put it into case text, scans, roll like cases, read them quickly.

within like seven minutes I'd have an answer, which now with ChatGPC and Claude, you can get, or all the other tools that are available like is table sticks, right?

Anyone can have it for essentially free.

But back in the day, back in the day as an eternity ago, five years ago, that was awesome.

And then I like became their number one user.

So Pablo reached out to me, like, hey, thanks so much.

Can we show you some new features?

And anyway, we hit it off.

I was one of the earlier users, co-counsel, it absolutely blew my mind.

um Unfortunately, I did leave Quinn Emmanuel to go start my first company.

And at that time, I had an offer to join Case Text about three months before they were acquired.

And they accelerated everyone's equity.

So one of the many poor financial decisions I had made in my life.

But Pablo's inspiration is just his passion for the project and him and Jake and just being able to carve something out of nothing was utterly inspiring and is one of the causes for necessary conditions for me doing for what I'm doing here.

So instead of sitting on a beach in the Caribbean somewhere right now, you're here talking to me about your next startup.

Yeah.

So that next startup uh is Learned Hand.

You're the founder and CEO and a whole lot of stuff I want to talk about, including your recent news this week.

But just for the advantage of people tuning in here who have never heard of Learned Hand or Slow Mo Klapper, what do you do?

Learned Hand is building a reasoning engine for courts.

Courts face a tremendous amount of volume of cases.

And in every case, their litigants are very well counseled, they're very slippery briefs, often lots of cases, people are not represented by lawyers at all, and that has its own challenges.

And judges face, and I can tell more stories about my own personal time as a law clerk, but judges face enormous responsibility.

and it's just this paper blizzard of cases.

And by the way, 90 % of judges don't have a clerk, dedicated clerk, so they're doing all this work themselves.

So what we're building is, I like to call it a judicial sous chef.

We're not replacing the judgment part.

I've never been elected to anything, not even student government.

I lost my bitch for student government.

I'm completely unelectable.

So we're not trying to take the decisions.

What we are trying to do is help them organize the information instead of them needing to go and search all the information and gather it.

They can, it can come to them and then they can spend more time in judge work and less time in drudge work.

One additional thing I will say is that you've had, there are lots of companies, many which you've profiled hundreds more that are, I feel like by the time we finished this conversation, there will be three legal tech startups that are being born.

But There's something called Jevons paradox, which VCs talk about, it's people talk about, but the second order effect of that is that litigation prices are coming down dramatically.

People are thinking about how that relates to billable hour, et cetera, but everyone's seeing that there's a secular shift towards that.

But that means cases are going to go up.

goes down, quantity goes up.

What does that mean for courts?

Courts have a fixed number of judges and they can't turn away cases.

So they will need tools like ours to help them more with the same resources.

Yeah, court AI is enabling courts to get not enabling courts is enabling litigants to flood courts with cases, sometimes legitimate cases, sometimes frivolous cases.

There was just a long article I was reading yesterday, Futurism or something like that was the was the publication, but it just popped up in my newsfeed.

But it was all about, you know, a couple of sort of significant but very frivolous cases that are that we're just consuming thousands and thousands of dollars and how, you know, there was a time when when Those either wouldn't have been filed or would have been disposed of much more quickly, but those times are changing and that's putting a lot of burden on the courts.

We say in the Pledge of Allegiance, of liberty and justice for all.

So the idea of access to justice is not simply a new phenomenon.

This is something that is core to the Republic.

And the idea is that we're a nation under laws and everybody should have access to this.

I was a big statutory interpretationer and I wrote several academic articles.

In terms of interpreting statutes, but the core idea of the dominant form of interpreting statutes is textualism.

Why?

Because the idea is that we should all that the statute should be accessible to the ordinary person, by the way, which is a whole lot of false assumption, but it's very normatively a nice idea.

So I actually think that having law be accessible, be cheap is a good thing, but we have to adjust the system for liberty and justice for all.

The answer is not as Sidley Austin is, is trying to sue open AI.

That's a frivolous lawsuit when Sidley Austin is suing and Nippon are suing open AI for unauthorized practice of law.

That's it.

That's the frivolous lawsuit.

Many people, are certain amount in which will, frivolous lawsuits will already be brought.

Previously, they were easier to do because they're handwritten and things, and now they kind of look like real.

So it takes longer to debunk untrue things.

But as Jared Perlow and NBC has written, there's been a lot of cases where people fought back against larger powers, right?

So people who have, like getting a lawyer for less than a half a million dollars, for a case less than half a million dollars is impossible.

But that can be life changing for people.

People can stop being evicted from their homes.

People can fight for custody of their children.

These are life changing amounts.

so I think that I actually have it contrary view.

actually think that there's a lot of good for having more access to the system, even despite all these things.

But we need to be able to increase capacity on the court side because otherwise you have bots versus bots and the humans are going to caught in the middle.

And when once courts start to break down, then it's hard to get their trust back.

Yeah, I'm with you.

I didn't mean to in any way.

I'm not all for uh litigants getting access to tools that are going to give them better access to the courts.

But if they're getting those tools, then it follows.

Go ahead.

Sidley Austin isn't, right?

found the law.

But like lawyers aren't, the guild isn't.

Yeah.

Yeah.

But if that's happening, then judges need access to the same tools, right?

I mean, that's the only way they're going to be able to keep up with all of that.

Yes, I think we need access to the tools for lawyers and the tools for judges.

Actually, I think are quite different.

They look the same, right?

Because they both generate things.

They both do research, analysis, drafting, but actually think in many fundamental ways are quite different.

I'm happy to talk more about that or oh I've taken us on enough time.

many things I want to talk about, I think just before we get to all of it, so we mentioned LA, so you're well, and we're going to talk more about this too.

But the news this week was that you've just done this, just entered into this partnership with the LA Superior Court system, which is the largest trial court system in the country to to do a pilot of your technology with the civil court judges there.

And it was an article.

in the LA Times this week about that pilot.

something that jumped out at me was this quote from the LA County District Attorney who uh in the article expressed concern about judges, called it problematic that judges would be using generative AI to generate their opinions.

And I'm thinking that a lot of people listening to this right now, especially because I have a lot of lawyers who listen to this, probably thinking exactly the same thing when they hear initially about what it is you do.

ah So are you replacing judge's decision making?

So you know, there's the short answer is, is no, it would be it would be foolish from an economic perspective, to try to sell judges a tool that replaces them.

I don't know if you've ever met how many judges you've met, but several hundred in the last two years, they are very skeptical bunch.

professionally, they're actually skeptics.

So they're they're very fiercely protective of their craft.

So they're, I think, There are two levels to this, Bob.

The first level is that is the sort of like an emotional thing that actually is important.

When people feel with a court system, they want to be heard.

They want to be heard.

And I think that's super important.

I think now the question is compared to what I think there are serious problems where cases like there is there, if I calculated the numbers once, if a judge were to read every single thing and every single case that were filed.

It'd have to be like six times the fastest reader in the world and they would have no, nothing else to do.

Right.

So there's simply not enough ways, time to read everything.

Um, I think there is a, there's a deeper question though.

Um, it's a steel man that DA's discomfort and the DA like the anonymous judge in the article, didn't actually didn't use the tool.

Didn't know, didn't really have a lot of insight into what we're doing, but there is a discomfort.

There was a white in case study which said, Lawyers approve of lawyers using AI 90%.

90 % of them are happy with it.

And lawyers approving of judges using AI, like 24%.

They're like, generative AI for me, but not for thee.

And the EU Act treats adjudication as a high risk category rather than law practice, not a high risk category.

The question is, what's the difference?

Why is adjudication?

different from legal practice.

And I think that the answer reveals why we need a different set of tools for judges and why we need a different set of tools for lawyers.

And I think that the main thing is that a lot of us use generative AI, but we know that it's really good at being psychophantic.

It's really good.

You can have two sessions of Claude put in precisely the opposite ideas.

And it would say that's Both of those are great ideas, right?

Those are fantastic.

And for lawyers who are zealous advocates, in many ways, this is a natural fit because your goal is to bend the truth as much as you can within your professional responsibility towards your client.

For judges, this is anathema because the judge's goal is to find the truth, to figure out what's actually happening.

So what people's concern is with gendered V.I., think it's just like chat GPD.

They put it in and you kind of get a magic eight ball and you're not sure what comes out.

So there's two levels.

Number one is we're not actually making the decision, right?

We are in essence like an AI law clerk.

I was a law clerk, a very good law clerk is structuring the information for the judge in a neutral way and assisting the judge.

They're not replacing the judge.

Again, like a matter of political theory and legitimacy, nobody's elected me.

Nobody elected.

Tech companies should be to have a little more humility, right?

We shouldn't be telling elected.

officials what to do, whether in courts or in the Pentagon, in either situation.

The other thing, though, is that we've really built this so that it fits with the judicial role.

That it is, when there is differences in facts, it will flag the gaps.

It doubts itself.

It will say, I generated this, but I'm missing x or y documents.

You might want to look into this.

Here are the gaps of the analysis.

There's an AI that doubts.

Like in Conclave, Ray Fine wanted the pope who doubts.

So this is the artificial intelligence that doubts the artificial, the AI that is built towards neutrality.

And I'll close with one story that, or study that illustrates this, is that one of our core partners did a study on bias.

And there's a very famous study in the employment context where you would take a resume and you'd set up a brochure, you'd swap the names, genders, ethnicities, same resume, right?

Harvard 3.8 GPA, worked at a research lab at MIT.

You'd send it out to employers.

And what would you get as difference?

You get a massive difference in response.

Not a little difference, a massive difference response.

Based on gender or the name alone?

is the name alone, whether it's a, not even the university difference, same name, you change the name and the massive difference because humans are not machines, right?

We're biased.

When you look at courts with same facts, same cases, but different backgrounds, there's not as bad of a spread, but there is still a gap.

There's bias that those have been studies that unfortunately judges are also humans.

The nice thing about a machine is that you can audit it.

So that's what they did.

You can audit a machine, whereas humans, can't really audit.

They did the same thing.

They took a bunch of cases, they swapped out the names and genders, et cetera, and then they threw it through the machine, our machine at learn hand.

And if it was just a correlation machine, it would have just correlated with past results, but it didn't.

There was actually no evidence of bias.

And the reason is that it's thinking mechanistically but like a lawyer.

It's going from the facts to the issues, from the issues to the rules, to the rules to the application, the application to a conclusion.

So I'm not saying that this is something that's going to replace judges.

I am saying though, that it can help not just make things more efficient, it can also assist with making things fairer.

And that's really what you need to build with judges is that there is reason why the DA is uncomfortable with AI.

The reason why white and case is uncomfortable with AI is because judges have a unique role and therefore they need specialized tools.

And those are the tools that we're building here at learning.

Yeah.

Jumping around the end a little bit, mean, you've had a somewhat eclectic career up until now.

You were a Penn Wharton undergrad, Yale Law School.

You did, in fact, clerk at the Second Circuit Court of Appeals.

You worked at Palantir for a period of time.

How is it that you came to want to focus on this kind of technology and this issue in the world of legal technology, given the various ways you could have gone?

It's almost as if my entire life is leading to this point.

There's not been a straight line in anything I've done.

And then I look around after 15 years and I'm like, holy smokes, like this is actually the only thing I'm put on earth to do.

it life is the universe is is odd like that.

each is, like bringing us together on a Manhattan street.

There you go, fate is doing it.

I think the story starts with my time as a law clerk, which by the way, was probably the most rewarding professional experience of my career.

So was right after I graduated from Yale, my first job was clerking on the Second Circuit, which is the court right below the US Supreme Court.

And my very first case was reviewing the sentence of first time non-offender who is 26 years old, who was sentenced to 30 years in prison.

And this federal sentence is really harsh, but...

I had a sense that nobody had time to really look into her case properly.

defense attorney certainly didn't have time, the sentencing judge didn't have time, but because it was my first case, I had nothing but time.

So I spent an entire week on the case.

I read all the records, I read all the case law, I read all...

I found many inconsistencies, and based on my work, her sentence was correctly reduced by 10 years, which was the best thing I've ever done for her.

But unfortunately, I couldn't do that for every single case.

because cases can be fast and furious.

And I don't think we got to any of the wrong conclusions, but the truth is, that had she come at the end of my term, she would have been serving those extra 10 years.

And so I asked myself, is this how we meter justice in the greatest country on earth?

And the answer is no, because it's way worse.

Most judges don't have the assistance of clerks, and if they do, they're often not at the level that the judges need.

uh State courts, where 99%, 98, 99 % of litigation happens, that was a federal public order, relatively well resourced.

Most litigations...

a relatively leisurely pace in a sense compared to a trial court.

Correct correct.

There's no statutory clock.

I had a case that sat on my desk for, I mean, it was actually for eight months because I just didn't have seven hours to write a first draft.

Now it's kind of hilarious that now with our tool can, and I mean, that's why we built this tool is I built this tool so that if I was a law clerk, I would have been able to give the same attention to every case that I gave to that first case and to make sure that justice is done in the time that's available so that When I was a lot, you talked about the statutory clock.

30 out of 50 states have clocks for judges in the appellate court.

We didn't cases.

Uh, there's one case actually that my judge dissented on, but he was, it was talking about due process and it was heard during my term and it just came out last year.

So it's like four years, a like it was just four years.

was like this case about due process.

I'm like, what world are you living in that you think you get into process by just those four years people aren't going to get back.

But there was one case that I actually, you know, I just sitting on until the end of the term because I just didn't have seven straight hours.

But if I had this tool, I'd be able to get a first draft really quick.

I knew what I needed to say.

It's just writing the section of facts, writing the analysis.

That was the hard part.

That was the judgework.

So if I had this tool, I'd be able to really get my arms around every single case, dig in, actually have a partner to think through things.

Because honestly, being a clerk, being a judge is very lonely because you are having a tremendous amount of weight on your shoulders.

And that's essentially what Learned Hand is.

It's essentially an AI clerk that can assist either clerks or staff, attorneys or judges do a faster, more efficient, but also deeper and more thorough analysis of every case.

In researching you, came across an article you recently wrote, a book review actually, where you kind of talked about your time at Palantir and kind of brought that forward to what you're doing now.

there was a quote in that article in which you talked about the fact that you are creating technology that respects the system it serves, again, harkening back to Palantir.

Can you kind of unpack that a little bit, explain what you mean by that?

Sure.

I think that one of the questions I often get is, they're kind of astounded that we're able to take an AI system and get judges to use it, and not just for peripheral stuff for what judges, but a system in the adjudicative function.

Again, in a very humble way.

And I think one of the keys of doing that is that word is humility.

I'm somebody who reveres the judicial system.

I was a law clerk.

I was a litigator.

I'm a lawyer.

I think in terms of what James Carville said in the Clinton era, he said, nothing is wrong with America that can't be fixed with the rights of America.

There's a lot of criticism with the judicial system, that the legitimacy of the system has gone from 59 % in 2020 and it's now like 34%, 35%.

It's like in free fall.

And I think that there's nothing wrong with the courts that can't be fixed with the rights of the courts.

I you have a cadre of some of the smartest, hardest working, most dedicated lawyers.

as judges who take their roles extremely seriously, all the staff attorneys they've met also take their roles extremely seriously.

And I think I'm coming in, I think people recognize when I come in with a sense of humility, I'm saying I'm not here to disrupt, I'm here to build and build together.

And I think in many senses that has roots in my time in Palantir with their work with the intelligence community, et cetera.

Palantir hasn't been able to penetrate the courts.

are really the first technology company, like modern technology company, to earn the trust of the courts.

And it's something that is uh a tremendous privilege and also an enormous responsibility.

Yeah, I mean, you are one of a kind.

We you and I chatted earlier this week before in advance of this call.

And uh I mean, I really I can't think of anybody.

And I think you said there is nobody else really who's doing what you are doing, specifically focused on the courts.

And so far, if I have this right, I mean, you've got pilots with 10 different state court systems. And is the L.A.

court number 11 uh on that list?

But I can't help.

But think, I mean, you mentioned earlier the fact that the judges are very traditional, they're very skeptical, they're supposed to be, that's their job.

What are the challenges of getting judges to want to even think about adopting this technology, let alone adopting it?

I actually think that judges are very skeptical, very curious.

And it's a really great interplay.

There's this organization called Jake, which is by three judges who are very pro-AI.

And honestly, they've had hundreds of judges sign up.

Judge Schlegel uh you guys in Judge Brotswell.

And they've been all been pretty vocal about AI, but there have been hundreds of judges that have signed up.

And they're keeping it, think, properly so very judge only.

There's no vendors, which is fine for me.

if there was, then my much better funded competitors, which are really Westall and Lex, would capture it.

the big challenge has been, I think there have been two challenges.

think the number one has been sort of coming in and saying, It's almost uh a pendulum.

It's like, is this worth anything?

Right?

And we have to show that we can deeply verify every single fact that we have a chain of provenance that it double checks it and that it actually can save you time.

It's both safe and effective.

And then the pendulum swings.

You're like, well, if it's so good, is it, am I unnecessary?

Right?

And so it's sort of navigating the green skill and sure.

And initially most of my work has been on the first side, which is getting people to see that this is useful.

And now more actually, more recently, the last few weeks, people with the demos that were doing people are like, oh man, this is, this is clearly excellent as like what role is there for me?

And I tell them is that nobody elected me, nobody pointed us like this is, this is supposed to be an assistant and, and truly a structure that way as well.

Like, I mean, one, one court that's trying us out, they requested like a no recommendation feature.

Like, great, that's fine.

My job is not to Like if you want to just the analysis and the research and the drafting, et cetera, and you don't want us to give a recommendation, that's totally okay.

Our goal is to assist judges with what they're doing, not as, they know what they need to do.

They just need the space and the bandwidth to be able to do it.

Yeah, we haven't actually talked about what it is you actually do.

I what does your platform do for courts?

uh it's a workbench.

It's a series of tools that assist judges from filing to the first draft.

And again, not actually, there's actually a range of judges where some judges are like, give me the whole rigmarole.

They want the whole thing, AI max.

Some people say, well, I just want the summary and the querying and the research.

And some people just want the hyperlinks of are we able to hyperlink?

So that's fine.

Let me, let me, I guess I gotta walk through the, series of, of a case of how it would work.

And then I can walk through a summary judgment motion of how a summary judgment motion currently works and then how we'd be able to assist with it.

with, um, with the platform, the way it's structured is, that there's different steps of a life of a case here.

So the first one, the first one is when a case comes in, you generally get, I don't have any printed out briefs here, but there's, get it.

This really thick amount of, reading materials, which are very densely worded.

But it's the briefs and then the records, and then they're very densely worded.

And they are talking as if you already know what the story is.

And for litigants, it's actually every case is their life, but for a judge, it's like 20 minutes on a Tuesday.

So I think the first thing we try to do is help orient the judge to the case.

One of the ways in which we do that is we're able to take the briefs and hyperlink them to the record.

Now this seems very straightforward.

It's not, it's not, it's actually really eye-opening for judges because instead of needing to open the record and et cetera, it's actually a difficult machine learning problem also.

like how do you have like Ambrosia, DECLA, paragraph 32, it's in a thousand pages.

Like actually fine, building the search operas was not that hard, but once you do it, it's really very eye-opening for judges.

After that, what we'll do is we'll read the targeted source and we'll let you know.

whether the lawyers are bending the truth.

Because as we said, mean, part of the art of the craft of lawyering is smuggling in things, facts and law, distinguishing cases.

You're trying to smuggle that in.

And there's this inequity here where lawyers have all this resources and time.

a clerk, I spent months on a brief that as a clerk, I would just scan, And then the judges are this.

So this is trying to give This lawyer BS detector is trying to give uh judges a little bit more uh equality in just handling the briefs.

Then there's once you have all the records hyperlinked, it's flagged, you can query it, right?

You can search it, which is we have like pinpoint citations, a very high accuracy of retrieval via legal research.

partner with a company called Midpage, which I absolutely love.

We have access to statutes for open laws, which is another company that I love.

uh and we are able to integrate those research sources in.

it's like really a one-stop shop.

then there becomes then, so until now actually most judges are actually like comfortable, right?

Everyone loves hyperlinks, everyone loves counter the lawyer BS detector.

Most people love the career, most people love them because it's just like retrieving information.

When you get to the parts which are harder to do accurately, ultimately more impactful, but.

will take a little bit longer for old judges to adopt, which is fine because I'm just trying to build tools to help judges where they are.

Then you come to the analysis part, which I love ah because I think it's the machines are so powerful at this and it can really help sharpen thinking.

So we're essentially for a given case, would run through it like a clerk would do it, which is where you have the, it would identify the issues, you would apply the rules.

It would essentially prepare a bench memo or draft an order.

well.

it would take and again like we're in a world is always compared to what some people judges or say like whoa that's scary that's weird and many of the judges who I speak to love this.

Why?

It's because they currently have to go into hearings cold.

Like one judge who's in a I think we're gonna convert them into full pilot soon, starting out as a full pilot into a full customer soon, is they told me when I first met him he's like my dream, my dream is to have a bench memo before going into hearing.

I'm like what?

I came in as a second circuit clerk.

I prepare bench rooms for my judge all the time.

That's what you do.

prepare bench room, read the briefs.

It's a workup of the case.

uh It's like, we can create that for you and we have.

And the ability to, instead of going into hearing just trying to rely on what the lawyers do to have a neutral aid in that and also to be able to draft an order, because honestly our hero judge is like a judge drowning in motions practice.

A lot of the time you need to just get your work done.

writing your dissenting Korematsu, right?

This isn't like you're not writing like, oh you know, the US versus Texas, right?

This is not like, these are not like Carolyn products.

These are just trying to help people resolve their disputes and they need to just get their work done.

And there are just so many cases.

So I think that it's really in that last two parts, the analysis and the drafting that people get uncomfortable.

I understand why.

And honestly, for judges don't want to do it, they don't need to do it.

We've got plenty of other tools.

I'm a Marxist in this way.

in a gradual Marxist.

Here are my values.

If you don't like them, I've got others.

And I say, are my tools.

you don't like them, I've got others.

You know, you talked earlier about dangers of bias, and I'm sure that something you hear a lot is that having this bench memo generated, this analysis generated is going to bias the judge toward whatever gets generated as opposed to somehow independently arriving at the same analysis.

You know, again, what do you say to that?

You know, I would say several things.

The first of which is, is that I think that it's simply not true to have a neutral analysis of a situation.

Well, like you have to have a very poor view of judges to think that they can just be swayed willy-nilly by the machine.

And the judges that I've spoken to, they have, not a single one of them has abdicated control to it.

And I'll tell you another story, Bob, about bias.

not all biases by the way, racial or ethnic bias, there's also bias.

I'll come clean here on the record.

until now, until about a few months ago, I was very biased against plaintiffs.

Even though I was a Quinn Emmanuel, which argues on both sides of the V, most of my work was on the defense side.

And I think that plaintiffs should just pound sand.

And they should just nobody should ever file a lawsuit ever.

It's a big tax on the system.

Let live and let live and let the market decide that not legal.

Okay so I exaggerate, but not greatly.

There was one case which was given to us by a judge in the earlier days of the company, earlier days when I was doing a lot more work.

Now I have really, really talented lawyers on my team who doing a lot of the quality assurance, but initially I was doing all of it, which is another reason why it's very hard to start this company, right?

Because you have to be a builder and a bit of a rule breaker, but also an institutionalist.

It's like a very weird set of capabilities that uh It's really, it's a blessing to be able to do it, but you have to have had a weird career like mine to do it.

And this was a case with South, Southwest was suing a major city because they were building a new airport, which I think is great.

More infrastructure, more public investment infrastructure.

Good thing.

And Southwest trying to have an injunction on it.

I'm like, plaintiff, bad, let the infrastructure go.

And essentially there were four slots in this new wing and Southwest was, there were five airlines.

And I said one.

airline needed to be booted from it.

And Southwest is suing that same thing that's unfair.

So I spent three hours reading this, three, four hours reading these briefs in this mindset of just like angry at Southwest and like, they're stupid company, why are they wasting judicial resources?

are they this?

They're just like reading through the cases and like, I just reading through the briefs or hundreds of pages of briefs.

Like it takes a long time to read it and read it well.

then you have to think much less hyperlink and run down all the citations.

For kicks, threw into Learned Hand and Ass for Bench Memo.

It kind of shocked me what came out because the benchmark was said actually Southwesters are really based on the evidence, based on the law and based on the circuit precedent, they actually have a really good claim.

And I was really shook because I thought I was doing a good job.

I was doing a horrible job.

That was a bad clerk because I was not coming in with the mindset.

But honestly, people are tired, people, et cetera.

The way you start something often determines how you finish it.

So to be able to have, again, this sort of another set of eyes, another another voice is not to say that it's going to take over because judges are judges because they're really opinionated and they're really skeptical.

I don't think so.

But I think what it does is that if I had that bench memo, I would have been able to have a neutral understanding, be able to probe the weaknesses and more quickly get into the depth of the case and also not have it out for one side, which when it's four, five o'clock on a Friday and then you get an emergency preliminary injunction, right?

you know, sometimes you're like, well, I just want to go home and get this over with.

So I think that the bias is, I think it's, I don't think that, look, I also, I also recognize that I might be wrong in this stand, especially for newer judges.

And that's why I'm actually working with one judge to build what I call hard mode or in the Hebrew raises Kavruta, which is study partner.

But essentially what it is, is that it's, it's not just giving you the finished product, but it's sort of making it intentionally.

sort of follow your own adventure where it's just raising the questions, forcing you to think, and having deliberate friction in the process.

So I'm aware that there is something.

think that the judges are already so overwhelmed that they want, they're skilled enough at what they do that that's not a risk, but I understand that some people are concerned, and that's why we're exploring building that for one court where they have the newer judges that have this intentionally hard one.

And again, that's not something any other vendor would do.

It's like we're all trying to eliminate friction.

And I'm sympathetic to the argument that in a frictionless society, there are costs to frictionlessness.

And what we might want to do is raise the right amount of friction.

So it's really a balancing act.

Yeah.

talking with Shlomo Klapper, founder and CEO of Learned Hand, about the AI workbench he's building for judges and how he's navigating concerns about bias in AI-generated analysis.

When we come back, we'll dig into the thorny issues of hallucinations, how Learned Hand won a head-to-head competition against a multi-billion dollar competitor, and its latest major announcement, a partnership with the largest trial court in the nation.

But first, please take this opportunity to learn about the sponsors who support this podcast.

Immigration attorneys, your time is too valuable to waste on outdated forms and manual workflows.

Lally Law delivers over 150 government forms updated within an hour of release, plus more than 50 customizable workflows designed specifically for immigration cases.

With real-time collaboration features and comprehensive case tracking, you can streamline every aspect of your cases from intake to completion.

Transform your practice efficiency today at lolaw.com.

That's L-O-L-L-Y-L-A-W.com.

Coming back to Law Next, I'm Bob Ambrogi and I'm speaking with Shlomo Klapper, the CEO and founder of Learned Hand, a company building AI tools specifically for judges.

Before the break, we talked about what the platform does and how Shlomo is addressing concerns about AI bias.

Now let's turn to another issue that looms large in any conversation about AI and the court's hallucinations.

Let's get back to the conversation.

The other issue that I'm sure looms large in any conversation you have about what you're doing is going to be that of hallucinations.

We all hear about AI generated hallucinations.

So why should judges, why should ultimately litigants not be concerned or should they be concerned about hallucinations in your platform?

Well, I think the first thing is, is that any gendered AI company should absolutely be paranoid about this.

And the time when it's most dangerous is when you're most complacent.

I'm embarrassed to say how much of our LLM costs it eats up.

The vast majority of our LLM costs are based on verification, not on generation.

And there are two ways of doing verification.

Because this is such an issue because in many ways Bob one of the reasons there's no competitors is not only because you have to have an awareness of the product The other thing is you have to have the credibility of building it.

It's tough to build but also because it's dangerous It's very high stakes.

I didn't if we have some major hallucinations, then that's really uh Then we're done meaning co-counsel be like Alexis all have hallucinations and they're all fine Because even though people have gotten fined for using them because they're big and they're not really, they're not going away.

For us, that's a very different story.

What we've, there are two types of ways of preventing hallucinase X, X anti and X post in what we're doing now.

What we have been doing recently has been a lot of very, a really brute force X post verification.

And we have quality verify.

Essentially what you do is you take each sentence and you treat it like interpreting a statute where you break up each clause and you're trying to see you're like, okay, here's the clause.

Here's the context.

And you basically run a thousand parallel, LLM calls, even multiple.

you're having the different models double check each one of those, it essentially it's brute forcing it, brute forcing checking whether that this particular fact is correct.

or not.

then you end up getting something that looks a lot like spell check, where it's just red, yellow, green, sort of this traffic light color to say like, well, this is okay and not.

What we're doing now is, and then the next step, sorry, I will finish that, is that you can then rerun it so that all the red stuff, you regenerate and then eventually it all turns green.

What we're doing now is actually, we found that the best levels of accuracy have been more programmatic.

For the lawyers listening, is that Large language models are probabilistic machines.

There's always randomness and in many ways that lends to their magic, right?

Is that they're flexible, but they're probably the most unreliable software ever created.

What lends itself much more to verification is relying on things that are code, regular expressions, identity recognitions, et cetera, a lot more of that on the ex-ante side.

where we're processing the information in such a way and then doing verification steps in all along the way.

in a very programmatic way so that by the time you end, you still rely on large English models for verification a little bit.

But the of the paradoxes of the LLM as a judge is how do you have something untrustworthy, checking something that's already untrustworthy?

You have like, you check it enough times, eventually it comes through.

But once we have program, we're building out programmatic verification, which is really, really successful.

Like our quotes, like we've been able to make it to that quotes, which are notoriously hallucinated by hallucinations.

Like I would bet.

I am betting the company that they're accurate, right?

And those we've been able to absolutely nail a lot of the basic, like the key facts we've been able to nail.

So there's just, it's not enough to have a closed universe.

It's just the first step of it.

What you need to do is have like extremely robust levels of verification, which by the way, might be, will be useful for lawyerly tools as well.

um I think everybody should be building with this level in mind.

ah but we were blessed that we've chosen the hardest customers.

So I want to ask you about the news this week of your partnership with the L.A.

Superior Court.

I know I had uh an opportunity once to meet David Slayton, the court's executive officer, get a tour of those courts.

And I know that uh he's been really uh focusing on doing any number of uh innovative partnerships and projects since he's been there.

But tell me about this.

me about what it is that your what the pilot looks like that you're going to be doing.

One note on David Saiten, the LA courts, David Saiten is the best court administrator in the United States.

When Bridget introduced me to David, she said that, and I kind of rolled my eyes.

It's been Bridget McCormick from the former Michigan Supreme Court Justice.

Yeah.

So, and now she's, you know, taken by the AAA.

So, you know, and I'm like, well, okay, that's very nice.

And then over the, wow, we've been working with each other for, geez, probably close to 14 months right now.

I've seen that firsthand.

The ability of him to drive innovation, to gather talented staff.

He's one of a kind in terms of court innovation.

And the leadership of the LA Superior Court is similarly...

extremely dedicated, as an extremely dedicated court towards innovation, towards partnerships, and ultimately towards serving their community.

The way that we started with the LA courts was, know, because as you said, judges are skeptical.

So they were curious because they actually handle more cases than I believe the entire federal court system in that one court.

Like they handle over a million cases alone in that court.

It's an enormous volume of cases, they're looking for ways of driving efficiencies.

As an aside, because I know you have a lot of technology companies that serve lawyers, one of the benefits of serving judges is that the incentives are completely aligned, is that they are trying to do their job in the best way.

They're not limited by any sort of billable hour.

Mm-hmm.

So what they did was, is that they were going to do an evaluation where they took their hardest motions.

And I insisted on the hardest motions.

said, well, do you want to start with easier motions?

I'm like, no, we start with the hardest motions.

do the hardest part first, it's easier from there.

And then we would run them through our tool and evaluate the output and say on a rubric of, you know, one to five on levels of quality of certain dimensions.

So you have like a very extensive multi-part rubric.

And essentially they would evaluate how good the output is.

um One of the large research companies, I'm not going say which one of those got wind of this.

And they said that their tool can do this also.

And I said, game on.

Let's have a head-to-head competition.

So we had a head-to-head competition with us and this multi-billion dollar company evaluating it.

And Bob, if it was close, I don't think they would have chosen the startup.

Even as innovative as they are, they wouldn't have uh have done that, but it wasn't close because for the reasons we talked about earlier, when you build for judges and you're building for high level work, it's very, very different from building for lawyers.

And it's very different from a general purpose LLM.

This is really a machine that can do legal analysis.

It's a very unique product as of now on the market.

And experts see the difference.

Experts can see the difference.

they decided, okay, now what we're going to do is we're going to test this so that it will scale.

So we say, okay, they had that initial evaluation.

Now we're going to have a certain evaluation groups about a dozen staff, where every dozen staff and six judges, six staff attorneys testing every going motion by motion.

Because judges think as trial court judges think in their world of motions and evaluating the quality.

And when it's ungated, they will release it to the literally hundreds of judges and staff attorneys that work in that court because it is such a large.

is larger than many state courts.

If LA was a court, like if that court alone is probably like 10th, know, fifth largest state, it's just a massive, massive court.

so they're having that test group.

And once we complete one motion, they'll do it.

And then we'll go motion by motion and starting with the hardest motions and eventually hopefully cover wall to wall.

There's also sort of, we started to speaking to other divisions within the LA Superior Court, which have own sets of problems, not the same as the civil division, but similar sets of issues.

Judges in the other courts we're working with, we work with family court, trial criminal, a little bit of probate, but they find that the platform is able to assist them with all their tasks.

And the reason is that judges everywhere are understaffed and overworked.

I'll talk about one particular thing for those who haven't been inside a courthouse.

I'll give one story about like a motion for summary judgment.

which if you speak to judges, as I have, go to judicial conferences, which exist, and you say, ask every judge, as I do, what is your most hated motion?

What is your most onerous motion?

And you have to say, like, obviously you have to love your cases, but what's the motion that wants to drive you into early retirement?

And Bob, across the board, every single judge, from Florida to New York, to Texas, to California.

Michigan, purple state, red state, blue state, says motion for summary judgment.

The reason is not hard to understand because a summary judgment is in a civil case after discovery before trial.

And one party says, your honor, here's all the facts we went through discovery.

There is no need to go to trial because I win.

And the other side will say, your honor, I completely agree.

There is no need to go to trial.

Here's all the facts, but I win.

And so what the judge has to do is presented with all these facts.

One party will say person X was the owner of said company and will give all the citations to the depositions, et cetera.

uh, the sources and the judge has to, again, open all the places and look through all the pieces and make it to zero K and keep that in mind.

And then you go to the other person says they're the owner and the operator of it.

And then they have.

20 more citations and you have to make a decision about something that seems like should be pretty obvious and then you have to do that hundreds and hundreds of times.

Most of that work isn't judge work, it is trudge work.

So what we do with the summary judgment motion is be able to pull all the information to the judge, have the statement of undisputed material facts hyperlinked and then what we'll do which I think is extremely useful, the judges think are useful, I can understand why people...

find it a little scary is what we'll do.

It will do an analysis of the disputedness and materiality.

But then you have to make sure that you link to the elements of the claim, right?

So that you're filtering whether it's material.

And then what you also have to do then is go to see if there's any evidentiary objections.

Meaning it's a very complex workflow that you can't vibe code.

You can't like vibe code this.

That's what I'm trying to say.

This is like immune to vibe coding.

But what it does is that it takes the responses we're getting from judges is that it would take things that would take them four hours and turn it to 10 minutes.

A motion that would take three hours and turn it to five minutes.

And that time is time they can better spend on hearings, on going deeper into a case, on getting cases done quicker.

Like this is time, like we are being able to give judges back their time, which is extraordinarily.

uh morning.

One of the features that's coming down the road, which I didn't talk about yet, is that we do the same sort of lawyer-based detector, which by the way is the same verification system that we do on our own.

This sort of post-facto, double-checking everything is the same thing that we use for the briefs.

And eventually is the thing that we're going to allow judges to use to review their own work.

So in many ways, there's two sets of, three ways of using Learned at Hand.

One way is extra information.

One way is as a first draft, but the other set is a second set of eyes.

I hope that gives you a better idea of what the...

Yeah yeah.

Correct me if I'm wrong, but I know the pilot in LA is civil cases only, and I believe your other pilots are civil cases only at this point.

Do you see this tool as extending into criminal cases?

And if so, are there, I don't know, constitutional issues there or other kinds of concerns about using AI in criminal cases?

So in the criminal context, actually think that there are certain risks, but I also think that the importance of this is even heightened.

criminal judges have most of what criminal judges do is trial work, right?

They're doing like learner in hand in many ways is the most helpful in the back office work.

which is a lot of cybill.

is a certain extent with other courts.

There are motions, so post-conviction relief is a big motion, suppression motions are uh big and those are ones we help with and that cleanly fits.

But a lot of what happens is you're in trial, right?

And then you're, like, it's not really that helpful to have, I guess you could have a little AI that tells whether uh hearsay objection is, et cetera, whatever.

Um, it's not, not really that helpful.

um so I think that the big question is about sentencing, which I think is extremely problematic for a machine to do sentencing and extremely imperative to machine to assist with sensitive meaning there's a tool called one of my bet.

Noah.

I have a few bet.

Noah.

So people who have systems that I can have evergreen like Daria Amode is one of my bet.

Noah's where I can basically anything he says, you know, is incorrect.

And then Compass, which is a system that's used for evaluating the risk of recidivism, uh is in my view extremely problematic because that's a machine making a decision essentially.

And even when it makes the decision, then it's like, and by the way, even when it makes the decision that a judge doesn't agree with, then you have a situation where, well, the machine said it, why would the judge, why the judge...

disagree with it, by the way, that is extremely, that's a very, very subtle thing that I'm afraid of in medicine, et cetera, right?

Where we ended up deferring to the machines and I will have to, if that comes up in our machine, we'll have to essentially make no recommendations um for that reason, because I'm very concerned about that.

But Compass, there was a public article in 2018 about this.

Essentially there's always a trade-off between type one errors and type two errors when you're black-rock prediction.

And you have to be honest about that.

You have to say, we're either like, we're either gonna put more innocent people in prison, we're let people who are not, are, you can't have it, or either you put away the innocent or you uh release the guilty, right?

Either you sort of, so that's a type one and type two errors, and they come at each other's expense, and you have to be honest about that.

And that's where you have these, machines are making decisions, and it's not auditable and it's a black box and you don't know why there might be reason.

It's a very, it's whole can of worms that's very messy.

On the other hand, there's been studies to show that when judges are hungry or upset or tired or in a fight with their spouse, I mean, we all make decisions that are enormous and we're all influenced by different things.

And by the way, judges make vastly different decisions.

like in asylum cases, it's not criminal sentencing, but asylum decisions.

are some cases, such as those who grant asylum 90 % of the time and some judges who grant it 1 % of the time.

There is no way that some judges get all the meritorious asylum cases and some get none.

It's just that judges have different views.

By the way, this is when the immigration courts work.

Now the immigration courts don't work, which is why we have this immigration crisis, why people are going extra legal things.

If you want to give an understanding of what, you want to give a glimpse of what happens when there's total courts breakdown.

looks at Minnesota.

That's what happens when there's no law.

There are your immigration laws on the books.

They aren't being enforced.

They're not being enforced by the courts, right?

Because the backlog is 3.8 million people, 3.7 million people, whatever.

It's a totally thing.

then we're resorting to this Hobbesian world of a war of all against all.

Point is, for a criminal case, I really think that there is something...

I wrote this article that I'll send you the link.

It is, I talk, I should have mentioned this earlier, but there is, it's a 200 something pages.

Nobody should read it except if you have sleeping troubles.

If you're sleeping troubles, we're going to submit for an FDA clinical trial that it should immediately help with insomnia.

But I distinguish between two types of uncertainty with judges.

And I should have mentioned this earlier about what the right role of AI is.

It's a difference between, there's some decisions that judges make, which is what I call epistemic uncertainty, where judges are just answering questions and nothing else.

other decisions which are constitutive, where there's no right answer, and the judge's decision is the right answer.

It's like collapsing a superposition of an electron in quantum physics.

There's no correct answer.

It's just the judge, by making the decision, that's the correct answer.

And that is a political decision that honestly has the violence of the state backing it up.

And there's no clearer place for that than sentencing.

When you're literally sending somebody prison, that is something where we have decided as a society, here's how we resolve these disputes and you have violence and we have not legitimated machines making those decisions.

So in my view is that I want the machines helping with the epistemic tasks and not making any constitutive decisions.

That being said, is that any intellectually honest judge will say is that they know that they are fall like, you know, Ulysses and they should be looking for assistance and help.

So instead of making a sensitive decision, how about a machine that would pull up similar cases?

How about a machine that will be able to look at prior and sort of be a devil's advocate, right?

We haven't built that yet, but I do think that even in these very, very sensitive uh situations, and I would argue especially in those sensitive decisions, even if we don't want the machine making the decision, I don't want the machine making that decision, but I would want the judges to have the best tools available to assist them in these crucial, crucial life and death situations.

Yeah, I know we just have a couple of minutes.

I don't know there's a short answer to this question, but I had wanted to ask you earlier and also forgot to ask you earlier, but what's the business model here?

I mean, so many of the startups I talked to, as we say, some of the companies I talked to, they're going after law firms, going after large law firms, big corporate legal departments, a lot of money, a lot of financing coming down, a lot of opportunity.

know, Harvey's raising money like crazy.

Legora is raising money like crazy.

How do you price this?

How do you raise money for this?

How do you build a business when the audience is a system that is perpetually underfunded and a little bit perhaps uh skeptical of what you're building.

Sure.

And I'll let you go.

audience actually really, they desire this.

Like we hear about, working with some major courts that I'd never pitched, but they've heard about us and they're like, I want this.

I haven't yet gone to a court presented what we're doing.

And then they say that's completely off base.

Like we're really hitting something.

think like we have spent a long time in the wilderness.

It took us a long time to get our first reports, but once we get there, like it's kind of.

We're, as a CHT company, very, very fortunate to get a lot of inbound.

What I'd say is that the business opportunity is actually enormous if you really understand this.

And I go back to my alma mater, which was Palantir, right?

The TAM for intelligence services in intelligence software in 2005 was like zero.

Maybe it was a few million dollars, right?

But nobody was selling, like maybe Microsoft was selling software to intelligence tools.

But by creating this tool, they were able to do what's called category creation.

They were able to create the category that then they dominate.

And there are certain elements that lead and category creation is like free solo in rock climbing, where a lot of VCs frankly are not looking to take risks.

They're looking to do things that are safe.

That's why they're all piling into the, nobody wanted to invest in Uber, everyone wants to invest in Lyft, everyone wants to be second.

And honestly, even though they claim to be risk capital, a lot of them are extremely extraordinarily risk-averse and treat early stage investment as late stage investment.

But there are certain investors who are very far sighted.

And investors who I found who joined, who funded us are actually the best in the business because the other ones don't get it.

And they see that.

They see two things.

Number one is that courts have, we spend an inordinate amount of money on courts.

We don't get a lot of results.

A tool that would allow us to get justice.

The budgets will be created for them.

That's the first thing.

The second thing is, It's clearly a winner take on market, right?

There will be, if we're able to get enough traction, there's going to be a thousand companies always trying to take the pieces of the advocacy side for the adjudication side because of the structural barriers, because I'm a one-of-one founder, because of the network effects, because it's a small enough market to dominate.

It will be, it was a winner, winner take on market.

And I would also say is that the market for adjudication is the Jevons paradox.

is going to grow, meaning the amount of cases we're going to see is going to explode.

And we can't increase the amount of judges on any reasonable time scale and at the exponential level is required.

And that will require, and we can't turn away cases.

So that means that everyone will have to, like once we have this tool that gives enormous levels of efficiency gains, state, the federal system will need to create the lines for this.

The interesting thing, the reason that why so much money is flowing to legal tech is that there are large language models.

are two areas of society that are completely bounded by law.

Sorry, by language, coding and law.

It's all just language, manipulation language.

And LLMs have, they've kind of cracked coding, they haven't yet cracked law.

I'd also say is that, so I think for the adjudication mark for this, this is not just a niche of like, let's say like for education, speaking to teachers where like, or even public defenders, right?

This is the core of the legal system.

there's, I'd also say is that, know, so anyway, so people, investors who are far side to be able to get this and there are category creations like Viva in the pharmaceutical space, like Salesforce, like HubSpot that once they create something ex nihilo are able to completely dominate it for decades.

And it takes a particular set of courageous set of VCs to do it, but those who get it are the ones we're going to look like geniuses in three years.

Yeah, okay my last question as you create this category, why did you choose to David after learning at hand?

ah learning to end with it, is the most famous judge to never reach the Supreme Court.

He's universally revered in legal circles.

I think actually the fact that he didn't...

Second Circuit judge where I clerked.

He's known by every law student and in many ways it exemplifies the approach to the business, right?

Which is we're coming from, like I'm not trying to say like Juris Bench or AI Juris or something.

No.

We are coming to restore the institution.

We are not coming to, this is a Birkin revolution.

This is not uh one of the French revolutions, right?

This is a conservative revolution.

are trying to, there's nothing wrong with the courts that can't be fixed with right with the courts.

We were trying to have an assistant to the judges.

And I'm trying to replace all the judges with AI.

I'm not trying to do this.

I'm trying to, trying to really give capacity to the courts because I love the courts and there's no better name of doing that than giving homage to the judge who not only lay the foundation of lot of American church students, but had the best name in courts.

I'll also say his brother's name was Augustus Noblehand, which is also, was also a second circuit judge at the time.

So, you know, the question really is what were their parents thinking?

Clearly, if they named them uh Harvey or something, maybe it would have gone different direction for them.

No offense to Harvey.

As one of my friends, Adam Hanford says, you can't exactly drive an Uber with a name like Lone in Hand.

It's either a home run or a strikeout.

It's also a good wink and a nod to the judges who understand it.

Every judge where get it, let them know that we respect them.

one of that.

We're here to help rather than just something that's a shiny new.

Silicon Valley and our next conversation, we can talk about how Silicon Valley has completely, why people shouldn't trust it and why we have, as tech people need to earn the trust rather than just assume it.

But that's a different, a whole other podcast.

Yeah.

Well, I'd love to do a whole other podcast.

We can get Pablo on along with you.

can have a nice light and wonderful little conversation.

ah Yeah.

Well, Pablo, if you're listening, thanks for suggesting I talk to Slo Mo.

Slo Mo, thanks so much for taking the time.

Thank you, Bob.

Thanks for this conversation.

Really a pleasure.

I'm looking forward to seeing you touch.

It's great.

Let's keep in touch.

Thanks for joining us for today's show.

I hope you enjoyed it.

If you'd like to share your own thoughts or comments, just go ahead and email me at broji.gmail.com or message me on LinkedIn or wherever you get your social media.

And if you're a fan of Law Next, please leave us a word wherever you get your podcasts.