COVID-19 Resources for Lawyers
Featured Guests
James M. Lee

James M. Lee is co-founder and CEO of LegalMation. He is a founding partner of LTL Attorneys LLP, a...

Thomas Suh

Thomas Suh is co-founder and COO of LegalMation. He brings a unique combination of legal and business experience to...

Your Host
Daniel Linna

Daniel W. Linna Jr. has a joint appointment at Northwestern Pritzker School of Law and McCormick School of Engineering...

Episode Notes

LegalMation has taught AI to speak legalese — how can lawyers use this in litigation processes? In this episode of Law Technology Now from Legalweek 2019, host Dan Linna talks to James Lee and Thomas Suh, co-founders of LegalMation, about how artificial intelligence can transform the practice of law. Their AI tackles the tedious tasks of litigation to free up attorneys for higher level work. They discuss the continuing growth of the system in more complicated areas of the law for innumerable applications in the legal industry.

Special thanks to our sponsor, Thomson Reuters.


Law Technology Now

Legalweek 2019 LegalMation’s AI for Litigation



Intro: You are listening to the Legal Talk Network.


Bob Ambrogi: Hello. I am Bob Ambrogi.

Monica Bay: And I am Monica Bay.

Bob Ambrogi: We have been writing about law and technology for more than 30 years.

Monica Bay: That’s right. During that time we have witnessed many changes and innovations.

Bob Ambrogi: Technology is improving the practice of law, helping lawyers deliver their services faster and cheaper…

Monica Bay: …which benefits not only lawyers and their clients, but everyone…

Bob Ambrogi: …and moves us closer to the goal of access to justice for all.

Monica Bay: Tune in every month as we explore the new legal technology and the people behind the tech…

Bob Ambrogi: …here on Law Technology Now.


Daniel Linna: Hello, this is Dan Linna. Welcome to Law Technology Now on the Legal Talk Network. My guests today are James Lee, Co-Founder and CEO of LegalMation and Thomas Suh, Co-Founder and COO at LegalMation.

James, Thomas, welcome to the show.

James Lee: Thanks Dan.

Thomas Suh: Thanks Dan. Thanks for having us.

Daniel Linna: Great to have you guys here. Well, we’re in New York at LegalWeek. Why don’t we just start with you guys telling us a little bit about your background as lawyers and journey to founding a legal tech startup? James, you want to kind of kick things off.

ames Lee: Yeah, so I started my career at Morgan Lewis and then at Quinn Emanuel, and at Quinn as fifth year lawyers, I and a few people decided to leave and start our own law firm, our boutique law firm. And we quickly realized that we didn’t actually know everything that we thought we did and we also realized that large firms were still going to bring in armies to these cases that we were on, and we needed to find better ways to litigate.

And so, we need to act as commandos and that meant that we needed to find ways to basically multiply the type of work that we were doing a technology was that solution. And so, over the years, we had grown our boutique firm to a pretty good size, about 40 lawyers, developed a good reputation and to carry that forward, that’s when we started to look at AI about three years ago and that’s how we started to dabble in it, really tested the limits of what AI can do and what it couldn’t do and LegalMation was a result.

Daniel Linna: And Thomas, so you guys met when you were at Quinn, is that right or tell us it kind of where how?

Thomas Suh: No, I’ve had the pleasure of knowing James prior to working together. We were friends socially and while he was at Quinn, this man always had the entrepreneurial spirit. While at Quinn, he was always thinking about new businesses and new ideas and almost told them you crazy. He’s just crazy. He is still crazy to this day.

And I think that’s what makes possible what we have done today, which is not afraid to try new things. So I did join him at LTL, the firm that he started about 10 years ago and we started working together with the purpose of growing the firm into a nationally recognized litigation boutique, and that’s what brought us together.

And this is just a continuation of that desire to build something from our experiences but really to make the practice of law more interesting.

Daniel Linna: Well, and you guys have gotten a fair amount of publicity, you’ve been written up quite a few times. I was very impressed when I first saw what you’re doing out at the Corporate Legal Operations Consortium Annual Meeting, but for our listeners who haven’t seen your product or heard about, can you tell us kind of what it does?

Thomas Suh: Go ahead. He always gives me the tough questions I got to say. So we’re a legal tech company certainly, but we’re solving specific pain points in litigation and what we did was our goal was to really tackle some of the volume process-driven sort of tedious tasks that involved in litigation particularly in the early stages of litigation.

And we set out not to just create a tool that will incrementally improve some efficiencies or incrementally help attorneys do their jobs a little bit faster, cheaper better. We really wanted to see if we can use AI to draft work product that associates and paralegals and other attorneys at firms actually create.

And so when we went back and looked at our own law firm and really looked at the early stages of litigation and what were some of the tedious volume-driven tasks that our associates were performing, which they didn’t enjoy, they found very — in the words of one of the reporters that covered us very soul-crushing, that’s what we decided to focus on.

And today, what you have is with our system you can literally automate the drafting of answers to complaints, outgoing written discovery, for instance, interrogatories and requests for production. And also you’re able to respond to discovery requests at least you’re able to — we’re able to create a shell to the response with targeted objections to those requests.


So what we’ve done is we’ve created essentially a new category and that is the category of automating actual litigation work product.

Daniel Linna: All right, so when I’ve told a couple of law firms with us, this guy is a former litigator and I was actually — in a lot of ways, I’m surprised that it’s taken this long for someone to create a product like what you guys created. So I think I see power and in a lot of different ways. I’ve had a couple attorneys say to me, they are like, oh, we have templates and we are pretty efficient at doing this. I mean, why is this better than having a template to do these tasks?

James Lee: Because a template is not a smart and dynamic system and that’s the big difference between template-driven solutions and what we’ve developed which is an AI Machine Learning-based system. So, think of it as the other people might have a TurboTax type of approach, where they ask questions and then it leads into the next sort of a phase of the question and then another question and then at some point, it delivers some output.

Our system, we wanted to design a system end-to-end where you can just simply upload a complaint and two minutes later, you get the finished work product on par of what a good first or second year lawyer would be doing.

Daniel Linna: I want to keep diving a little bit deeper in this, right, because I think one of the challenges we have in the legal industry now is it’s really easy to say that you’re using Artificial Intelligence and then no one really asked questions. They don’t know how to ask questions maybe even.

So — and you mentioned Machine Learning and tell us a little bit more about the Machine Learning approach that you are using.

James Lee: Well in various conferences, the way I explained Artificial Intelligence is simply — think of it as a pattern recognition machine, that’s all it is. The more patterns you can show it, the more examples it’s going to find patterns within the examples that you give it.

And the way it works is you can think of it almost like a coin sorter where you take unstructured information whether you have a bunch of coins, you put it through a hopper, and what the AI platform does, it starts looking through and sorting, very quickly in this case words or phrases to put it into neat little sort of data fields or entity relationship pairs.

And we take that information once it’s been classified and then we start refining it even further using several different techniques to essentially reach the same conclusions a lawyer might be reaching. And so what we’re trying to do is we’re mimicking what a good first or second year litigator might be doing under the similar circumstances.

Daniel Linna: And what platform are you using for Machine Learning?

James Lee: So we use a few actually. We use IBM Watson for some of the functions and we have built a couple of other AI classifiers on top of that to refine the decision-making. So what we’re finding is that one AI classifier isn’t the best for everything and the way we sort information is actually quite strenuous. We take several different approaches to take a look at the same sentence or a paragraph in order for us to determine what is the best way to sort of parse that paragraph and then to obtain the information that we want.

So sometimes Watson works great and if it doesn’t, then we’re using a second classifier; if that doesn’t work, we’re going to use a third classifier using different well-known techniques in the Machine Learning world.

Daniel Linna: Sometimes when I hear some of these problems that were approaching discussed, it’s framed just as a data problem, so there’s a lot of variance across attorneys and I’ve heard some people say, well, that means we’re going to need a lot of data. I mean, I wonder actually if those approaches are going to work very well because there are some areas where we don’t seem to have standards best practices and if you can just keep adding more data and there’s tons of noise and variance in the data, I mean I don’t know where that’s going to get you.

How do you guys kind of solve for that in your solution here? I mean, how much kind of expert input is required versus just looking across the documents?

James Lee: I’ll handle that, Thomas.

Thomas Suh: Thank you.

James Lee: I think the best answer to that is it depends and what I mean by that is it depends on the domain, it depends on what you’re asking the system to do, it depends on whether the likelihood of ambiguities exist. The likelihood that there’s more ambiguities means that you have to give it more examples to identify the different classifications that you want to teach it.

I think the mistake that a lot of people are making in the AI world and you read articles in the newspaper about medical institutions that have spent millions of dollars and it just failed. I think what’s going on there is they were probably using an unsupervised learning process, probably led to believe that you can just feed tons of data in the form of medical journals and somehow it’s going to magically learn medicine that’s not how it works.

You frequently do have to use a supervised learning process particularly for what I’m going to call professional languages. Legalese is one of those languages, it’s not normal English. You cannot use a standard AI dictionary that can parse that natural language, you have to teach a system what a plaintiff is and all the variations, what a defendant is and all the variations, and all the ways a defendant has been accused of screwing off for a plaintiff.


All those classifications you have to think of and we’re not talking about 10 or 20, I mean, we’re talking about several hundred that you have to think about and then think about the combinations of those pairs in order to come up with the right ontology or the classification scheme that you’re going to get the best results possible.

Daniel Linna: And if I understand the way you guys built your system too, I want to keep drilling just a little bit deeper on this. It’s not like, if I told you, oh, well, let’s — we’re going to do this for a certain type of law in California and I’m going to give you every complaint and every answer. You guys would say, well, I don’t think I want every complaint and every answer. You are actually a bit more targeted, you want a specific type of information data to train your system, and again, I think this is something that’s missed frequent and people talk about this while, oh, you just need data, well, but what’s the quality of the data and —

James M. Lee: Right, so this is part-art and part-science. The art part is coming up with a good combination of cases to expose the Machine Learning platform to enough examples where it’s going to be able to find patterns, right? And so, our system is heavily curated. It’s curated by lawyers and what we do is we’re really looking at the samples, the type of cases that would be — that we think the system is going to encounter and what we’ll do is we’ll go source publicly file complaints and we’re going to start basically teaching the system, how to identify all the necessary claims.

It’s a curated process and iterative process where, for instance, there was a time when our Motor Vehicles domain didn’t do a good job identifying traffic violations and so what we did is we went out and we curated and found really good examples of traffic violation cases and then fed that into the platform and then we did the learning models and then tested and sure enough it was able then to then identify traffic violations, and so that’s a sort of iterative process that people have to go through.

It’s a long process. It’s not a magical process where you do it once and like I Dream of Jeannie, you blink your eyes and something comes out, it doesn’t work like that. It takes sometimes 20, 30, 40, 50 times before of constant testing, retesting, uploading, testing, retesting to see whether you’re getting the results that you’re getting.

Daniel Linna: And is this, I mean, the way you’ve deployed it in most places, is this like an online learning system? It continues to learn as you go on like is there a continuous training? How do you handle that part of it?

James M. Lee: Yeah, so that underscores the biggest problem in the AI Machine Learning world which is data bias. A few weeks ago we were listening to that lecture at Stanford about data bias and it’s the biggest problem and the way that we solved it is when clients run new complaints through our system it doesn’t affect the learning domain and the reason why we made it that way is, we didn’t want a client, or a law firm, or an insurance company to run a certain number of claims of a certain type to skew the results in any particular direction. So what we do is every couple of weeks we’re going to look at the results and see do we need to supplement a domain, are we missing anything. Let’s update the domain. Well, let’s find new samples, let’s put that in and so that’s the curation part that I’m telling you about is that we want to keep that pristine as possible so that we don’t end up with that data bias that I think some other companies are likely to experience.

Daniel Linna: In what jurisdictions now do you handle? What types of cases, I mean, where are you guys available at this point LegalMation?

Thomas Suh: Sure. So we started in California because we’re California lawyers and we practice in California.

Daniel Linna: Yeah.

Thomas Suh: And the office space we already had so that was one way to keep overhead low, but California and we have Texas, we have New Jersey and Florida is coming online in a couple of weeks, and shortly after that in New York. So we are targeting sort of the large states in terms of volume, not necessarily population. Maybe with the exception of New Jersey which we sort of moved up the ramp because of a large customer who wanted to get to New Jersey first.

In terms of practice domains, we started with employment litigation, employment cases. Then we moved on to all sorts of tort cases, so product liability, slip and fall-type cases. You’ve got — what else we have there and torts.

James M. Lee: All torts, all injury type cases, motor vehicles was included as well.

Thomas Suh: Yeah, pharmaceutical cases, pharma product liability and the third domain is insurance bad faith cases.

James M. Lee: And we have financial services that are — that’s an alpha mode right now that we expect to beta-test maybe in a few more weeks.

Thomas Suh: So essentially we’re targeting the large volume cases that really makes up for about 80% to 90% of the litigation docket in the country.


Daniel Linna: Okay, yeah, and it sounds like you guys are starting to tackle more and more complex areas as well because I know when there were first some stories it was heavily discussed about Walmart and handling cases for Walmart and a lot of lawyers said, oh, well, slip-and-fall cases but I don’t do slip-and fall-cases. I do much fancier things and so this isn’t going to affect me. I mean, where do you think — I mean how far can you keep going kind of with this model to automate portions of the litigation process, drafting documents and so on?

James M. Lee: I actually think we can go pretty far and the reason why I say that is, people think of us as a company that can produce answers and discovery but that’s not we really do. Our real value is the ability to take unstructured information in the form of the complaints or documents, use a Machine Learning AI process to structure it and then to classify it in many, many, many different ways and when I say “many, many, many” I’m talking about we are — each of our domains we probably measure about 500 entity relationships and if you were to look at the number of permutations, and the number of combinations you can get, at ten positions which actually isn’t that many, a plaintiff, defendant, claims, injury date. At ten positions that ends up being point 0.89 octillion number of possible combinations, that’s 10 to the 27th power. That number should scare everyone in this room because we are barely scratching the surface on what we can do with the system, and so you know what, the next products that we’re thinking about is to use that ability to take on structure and information and structure it in so many different ways that we can then deploy better analytics, case research even responses to simple motions, maybe even preparing motions to dismiss.

There’s so many things that we can do that it’s really exciting to think about. If we had a blue sky session, we can probably map out 20 things that we can do just by our ability to take unstructured data and then to structure it.

Daniel Linna: And who are some of your customers right now like we know Walmart because that’s been in the news and I know that there were some other who have been publicly disclosed. So can you disclose the ones you can and tell us maybe some of the other where you think you see opportunities?

Thomas Suh: Well, James is looking at me like I should answer that question, even though we’re subject to NDAs for a lot of them. The truth is we can’t disclose I would say a majority of our large clients but I will say, well, Ogletree is the one that we can Ogeltree Deakins has been a phenomenal partner for us. They are really led by Patrick DiDomenico, their CKO. They’ve been at the forefront of change and embracing technology for the practice of law and we’ve entered into a special partnership with them. So they’re going to assist us and we’re going to work together on a lot of things that probably can’t disclose right now, but that’s one company.

We actually serve very large insurance companies as well but we can’t name the names and a number of let’s just say, a great number of Fortune 100 companies are currently in pilot testing and about to roll things out, so we’re very excited.

Daniel Linna: So, tell me more about using this kind of like for early case assessments. That’s something we’ve chatted about before maybe. Is anyone really starting to try to make progress with that? Like really trying to build out a model for predicting where a case is going to go and being able to use the power to analyze a complaint when it comes in right from the beginning to —

Thomas Suh: Yeah, so nothing like that is available now, but that’s something we are going to be working on and that’s the real exciting part about all this. To just to give a concrete example, if you want analytics on a wrongful-termination case you can get that today. You can get a wrongful-termination report and it can give you ranges about how much you should spend, what the settlement value should be, et cetera. But experienced employment lawyers will tell you that all wrongful termination cases are not the same. People that have been employed for a number of years much different risk than someone who’s been employed for six weeks and there’s no way to capture that information efficiently.

Well, our AI platform can capture that information. Not only can it capture the wrongful termination case but if it’s in the complaint, figure out how many years he or she’s been employed, the basis for the termination, the injuries that he’s alleging, the amount that he’s alleging, the name of the plaintiff’s lawyer. All these factors that come into play, we can capture in seconds, and essentially then deploy that into a more comprehensive analytics program to give you a more nuanced view of the risk of your case. It’s almost like a digital DNA fingerprint of a complaint.

Daniel Linna: What about — how does the review, the work product work right now, like when you get a draft of the answer, I mean, how much surely the lawyers are reviewing the work product before you file it in the court? I mean, what does our review process look like? Are you guys trying to capture data about what kind of changes are made, how much change needs to be made before it can be filed?


Thomas Suh: I can try to take a stab at it. Somebody asked me that question earlier today. So we’re not tracking keystrokes, maybe that’s where you getting out and that’s one way to do it, which is, is someone editing the document online and can we see sort of the changes that are being made? No, right now, once the user runs the complaint through and they get the answer and the initial set of written discovery, it ends up on the other side of the fence.

So they will download it and whatever they edit through their basic Word — it’s a Word document, so whatever they edit we don’t see those changes. I can tell you this though from a constant feedback that we get from attorneys. I think the best measure is really the time that they spend to finalize the documents that we’re providing. In the early days when we were just ramping up with Walmart, for instance as an example, we knew that their outside counsel were spending maybe half an hour to an hour editing and finalizing the downloads, the documents that we produced, I recently contacted several of them, and now what’s happened is that editing, finalizing time has gone to about 15 minutes or less.

And the reason is quite simple because now they realized that the system is very consistent, they have gotten used to it now. They have learned to trust what is coming out of our system and so by that fact alone, we can tell you that they’re saving about 70% to 80% of time in drafting these documents.

Daniel Linna: Yeah.

James Lee: And when I was a partner, when I would wait to get work product from young associates, sometimes it was good, sometimes it was —

Thomas Suh: Watch your words, watch your words.

James Lee: I know. Sometimes it was terrible. And that —

Thomas Suh: Thank you.

James Lee: That lack of consistency was a big problem and that’s one thing that we were striving to solve is can we develop a platform that was consistent enough where the reviewing attorney can trust it, if there’s any weaknesses we’ll know where it is, where it strengths and then just move forward from that point and I think we’ve done that.

Daniel Linna: Yeah. Well, I think though it’s interesting that the way that’s framed because I think it also gets back to the training idea, right, because associates may be inconsistent but partners are inconsistent frequently as well, right? And maybe the work product that you thought was excellent, maybe Thomas wouldn’t think it was or someone else would want to do it differently, and I think one of the interesting things about these platforms is I think sometimes people worry about whether the machines will produce high enough quality, but whenever we are going to get to a point where people are going to say, look, the people shouldn’t be editing and changing this. The machine is the highest quality product we have.

Thomas Suh: Well, I don’t know if we ever want. So from a practical standpoint, from a technical standpoint I think that’s very possible. From a practical standpoint I think you always want the attorneys to sign off on the documents because obviously we have ethical rules and licensing rules and regulatory issues.

Daniel Linna: Yeah.

Thomas Suh: But you’re right. I mean, from a technical standpoint it’s possible and I think it probably will get there, but you always do want the human eye to sign off on it.

Daniel Linna: Sure, sure. Yeah. Well, and there’s a difference between reviewing it, like we supervise work that others have done and when we outsource work and things like that, the lawyers got the responsibility. But I mean to me, maybe the more important point to make rather than pushing the envelope on this, is that I think sometimes people think, well, this is all just about efficiency, right? This is just pushing, but quality, right, we’re getting higher quality when we use these tools as well.

Thomas Suh: Right. Right.

Daniel Linna: All right. Well, let’s take a quick break right now. So before we continue our interview with James Lee, Co-Founder and CEO at LegalMation and Thomas Suh, Co-Founder and COO at LegalMation. We are going to take a quick break for a message from our sponsor.


Monica Bay: Thomson Reuters’ Westlaw Edge is the most intelligent legal research platform ever. Powered by state-of-the-art Artificial Intelligence, Westlaw Edge delivers the fastest answers and the most valuable insights, providing you with a clear strategic advantage. The advanced features on Westlaw Edge allow legal professionals to practice with a greater degree of certainty and confidence, never before available. Visit to learn more.


Daniel Linna: And we are back. Thank you for joining us. We are with James Lee, Co-Founder and CEO at LegalMation and Thomas Suh, Co-Founder and COO at LegalMation.

So guys, we have been talking a lot here about how your tool automating different parts of the litigation processes, it’s going to have an impact. Well, why don’t you tell me more, I guess, so let’s say about the impact that you think that your tool is going to have on the way we litigate cases, way we formulate strategy, where lawyers going to really add value when they’re using these tools, what do you think?

Thomas Suh: Yeah, very briefly. I mean, the automation of the early stage discovery documents and answers and pleadings that we are currently doing already, they’re fantastic tools, if you will.


But they’re just a small sliver of what the real impact is on litigation strategy overall. And from our view and from the view of clients what this does now is it kind of shifts some of the — actually shifts a lot of the time, but is usually spent on grunt work and scut work and gives the time back to attorneys to focus on strategy, on investigations and other things like that.

In addition to that on the Discovery Request Module that we have, if you’re saving that much time in an activity that is typically very soul-crushing, if you will and repetitive and you’re saving that much time, the leverage that was used by, let’s say a large firm against a smaller firm in litigating a case where the term “burying somebody in paperwork” is no longer there. This levels the playing field.

So I think litigate strategy change because of that, it can change at many different levels as well with the other things that we have planned on the data analytics side and many other things. I think we can fundamentally change how lawyers view this tool to litigate their cases.

Daniel Linna: Well, just based on you guys have done already and what I’ve seen other stuff, I see 00:26:20 doing with judicata. I mean there’s a lot of people are doing stuff where I think mapping a loss, laying the foundation to automate even more processes. I mean, I don’t think — I think I can’t imagine we’re too far away being able to automate the drafting of a motion to dismiss and the brief that would support it. What is the — I mean, I think there’s still something there left for the lawyer to do. What does it look like for the lawyer? What’s the lawyer’s job when you have this first draft coming out of LegalMation of a brief supporting a motion to dismiss, for example?

James Lee: I am sorry, what was that?

Daniel Linna: Yeah. Well, I mean, like the way I envision it and sometimes I see, I see lawyers freelancing a little bit, are like kind of — we talk about specializing, right, but how many lawyers really, really know the industry, really that they’re working in really, really know their client, the judge, right? What if we had more time to really be laser focused? I mean, might that be a way that litigation changes?

James Lee: Yeah. I think once we can push the value up, further up the chain, the better. I think we will all start experiencing better outcome for clients. There have been many times and I’m sure with you too that you wished, you had more time to prepare for a deposition.

Daniel Linna: Yeah, yeah.

James Lee: You wished you had more time to prepare an opposition to a certain motion. You wished you had more time to prepare a cross-examination. Well, tools like this give lawyers that time so that they can start really practicing their craft to get better outcomes for clients.

Daniel Linna: Yeah. Yeah. One of the other things I hear about this space and we’re we are making predictions at this point, but I’ve heard people say, oh, there’s going to be more and more data and there won’t be any cases tried. On the other hand, I think, hmm, there are a lot of cases when I was litigating that, probably should have been tried, but it was just going to be too expensive to get it to the stage of trial. I mean, maybe this will help us figure out more quickly what cases ought to be tried and —

James Lee: Yeah, I think once you have more detailed analytics you will really get to see a real Moneyball approach to litigation. They use Moneyball in sports, and baseball, why aren’t we using in a litigation, right?

And so, I think you’re definitely going to see a day and that day is going to come fairly soon where clients are going to have more visibility and more transparency into the risk, more visibility and transparency into legal spend and how much they should be settling cases.

Daniel Linna: So, Artificial Intelligence, there’s been some hype, but some people then kind of write it off and where do you think are we going to see a lot bigger things than even just what we’re seeing right now, like what’s the impact going to be in the legal industry longer term do you think?

Thomas Suh: So I think, at LegalMation at least we — if you use Artificial Intelligence in the right circumstances for the right projects, for the right tasks, I think it’s a fantastic solution. I think the big mistake that people make is having the impression that it can do it all and eventually it’s going to be able to do it all.

I don’t know about that. I think it’s — maybe we’re far away from that sort of a — the idea of singularity or something of that sort or a general AI if you will. I’m not going to dismiss it offhand, but from what we have seen so far, it is fantastic at doing specific tasks in a big way and very fast.

And so if you know how to leverage that for certain tasks in the legal field, I think it has tremendous effects. I do think it is going to improve. I think it’s going to get probably faster, better and all that good stuff, but I’m not sure if it will have the impact of replacing an attorney’s strategic thinking on a case, or perhaps being able to appeal to a jury while they’re arguing in court, because those are sort of intangible, less data-driven and more field type of activities and may be AI will be able to do that someday, but I think we are far away from that.


Daniel Linna: What about — I know I’ve spoken with both you guys and you’re interested in possible applications and the access to justice space. Tell me where you think this could really have an impact?

Thomas Suh: Yeah, for a lot of us at the company, the promise of AI and legal AI is not only to do the work that we’re doing but also to spread those advantages to as many people as possible, and you know what, a lot of people don’t realize is that this technology can be used in other contexts, and one of those contexts could include access to justice.

What do I mean by that? Well, there’s a lot of pro bono organizations that are turning away people because they don’t have the resources. Well, if we can help them or if other AI companies can help pro bono organizations serve more people, that’s a good thing and that’s one of the things that we’re striving to do.

I mean we are talking to a couple of organizations that — where we can apply our learning to increase veterans’ benefits, landlord-tenant type of disputes, that brings a smile to all of our faces because that that’s an example of where I think AI can really be used for good.

Daniel Linna: James, you mentioned a lecture at Stanford that we were both at, I was at Stanford with J. Mundell and we put on a boot camp there for law, computer science and graduate business school students around exponential innovations, AI and law, getting them to think about learning about these tools so they can evaluate them, thinking about business models, thinking about the application of law and ethics in that space as well. And you were kind enough to come and spend a full day plus with us as a coach to those student teams. What do you guys think we ought to be doing in law school in light of the technologies in a way practice is changing?

James Lee: So, that’s a great question and there’s a couple law schools that we’re talking to right now. We’re talking about having some students help annotate. So training a Machine Learning platform, maybe even use that so that the students at the clinic start using that tool so that it’s a — sort of an ecosystem at particular law schools where they’re able to use AI, both from a practical sense as users but also as another practical sense where they begin to really understand and how to train an AI system. And from a training perspective, I think that’s fantastic.

Daniel Linna: What about just more holistically, I mean, do you think we need to be changing the way we train lawyers in law school in light of the way the marketplace is changing? I see you nodding, yes, Thomas. What do you think on that?

Thomas Suh: Yeah, I mean, like sort of like, I mean at least for me. My opinion is that there’s a place for — I think we need to shift — it’s sort of like these technical schools, right, not the general universities and liberal arts schools but more than technically very focused type of training and platforms that you have.

I think law schools need to shift to that. I think law schools for too long have been more caught up in the theory of things and the possibility of things and studying sort of the old cases, and that’s important stuff, but I think it’s got to become a practical; meaning maybe partner up with actual attorneys that are practicing and maybe increase that access, get them more hands-on training, right, so that they’re ready to roll as soon as they step out.

I mean, let’s face it, corporations are specifically telling their law firms not to put first-year associates on their cases because the typical first-year associate at least today — as of today, their value is limited because of what they know or really because of what they don’t know.

So I think if you create a program where the first year junior associates are out of law school but ready to roll into a litigation shop or whatever other firm but they’re able to produce and contribute meaningfully right off the bat, then I think that’s where the focus should be. And a lot of law schools are thinking that way. They’re becoming more training grounds as opposed to the theory academic arena.

Daniel Linna: Yeah, yeah, and then I look at the kind of products you guys are building and what I want to do like what we’re doing in Northwestern, I want to train the lawyers who are going to see the opportunities that folks like you can help them deliver on, but also know how to ask the tough questions to make sure they’re fulfilling their professional obligations, make sure these tools actually work the way they’re supposed to and they understand really the benefits and risks of using the tools.

All right, well, thank you very much for joining us, you guys. Can you just tell us, James, starting with you, about how listeners can reach you if they want to get in contact?

Thomas Suh: Yeah, give them your home address and your personal email, no I’m just kidding.

James Lee: Actually I have no ide.

Daniel Linna: Well, LegalMation, I know LegalMation is on Twitter. I mean you guys need some sales training I think, LegalMation is on Twitter.

James Lee: We are also as much as the president but we do our best. Listeners can reach us through our website. We have a contact form. We are happy to put them on our list and distribute as we had more news and certainly if there’s any interest out there to see a demo, they can certainly — we can put them in touch with one of our folks.


Daniel Linna: All right, well, I appreciate you guys joining us here in New York during LegalWeek. This has been another edition of Law Technology Now on the Legal Talk Network.

If you like what you heard today, please rate us in Apple Podcasts. Join us next time for another edition of Law Technology Now. I am Dan Linna, signing off.


Outro: If you would like more information about what you have heard today, please visit, subscribe via iTunes and RSS, find us on Twitter and Facebook or download our free Legal Talk Network App in Google Play and iTunes.

The views expressed by the participants of this program are their own and do not represent the views of nor are they endorsed by Legal Talk Network, its officers, directors, employees, agents, representatives, shareholders, and subsidiaries. None of the content should be considered legal advice. As always, consult a lawyer.


Brought to You by

Notify me when there’s a new episode!

Episode Details
Published: February 13, 2019
Podcast: Law Technology Now
Category: Legal Technology
Law Technology Now
Law Technology Now

Law Technology Now features key players, in the legal technology community, discussing the top trends and developments in the legal technology world.

Listen & Subscribe
Recent Episodes
Black Lawyers in Major American Law Firms: How to Make More Progress

Harvard’s David Wilkins and Host Ralph Baxter examine why law firms struggle to hire, retain, and promote black lawyers and how they can do...

Good Housekeeping Seal of Approval for AI?

To achieve wider adoption of AI tools, there needs to be more industry testing and vetting, Prof. Maura Grossman tells host Dan Linna.

Model for Change: Utah’s Data-Driven Approach to Closing the Justice Gap

Ralph Baxter hosts key players in Utah’s move to reshape the delivery of legal services, revealing the aha moment that sparked the movement.

Pros & Cons: Data Privacy’s Role in Advancing Legal Tech

Host Dan Rodriguez and German lawyer Markus Hartung parse the differences between legal tech advances in the U.S., U.K., and European Union.

Hotshot: 21st Century Training for New Lawyers and Law Students

Ralph Baxter hosts Hotshot co-founder Ian Nelson and Harvard’s Sara Dana and Morrison’s Rick Jenney to discuss how Hotshot’s videos teach practical skills lawyers...

The Spanish Flu to Covid-19: How this Pandemic is Pushing Courts to Modernize

Michigan Chief Justice Bridget Mary McCormack details how courts are breaking with century old processes and outdated technology to build trust and serve the...