# LexisNexis CEO says the AI law era is already here - **Show:** Decoder with Nilay Patel - **Published:** 27 Oct 2025 - **Source:** https://www.podtrac.com/pts/redirect.mp3/pdst.fm/e/pscrb.fm/rss/p/mgln.ai/e/257/traffic.megaphone.fm/VMP5540704411.mp3?updated=1761338807 - **Transcript:** https://podcasts.opennash.com/shows/decoder-with-nilay-patel/lexisnexis-ceo-says-the-ai-law-era-is-already-here-0b5b8e/ - **Provider:** OpenNash Podcasts (machine-generated) ## In brief LexisNexis CEO Sean Fitzpatrick argues that AI built on a closed, verified body of real law can give lawyers a trustworthy drafting and research assistant. He says general-purpose chatbots fail in court because they invent cases and cannot show their reasoning. LexisNexis says its tools link every citation to a real, checked case and keep a human lawyer in charge. The tension is between speed and the profession's future. Host Nilay Patel presses on who will train junior lawyers once AI does their work, and whether judges using AI will automate justice itself. The sharpest clash comes over originalist judges using AI to study historical word meaning. Fitzpatrick calls the tool neutral and declines to name any argument his AI would refuse to draft. ## Key insights 1. **Consumer chatbots fail courts on five specific counts** Fitzpatrick says general models use stale data, can make up cases, offer no attorney-client privacy and show no reasoning. LexisNexis counters with a curated 160-billion-document collection, a citator agent that checks cases, and visible logic. 2. **AI could break the junior-lawyer apprenticeship model** Both men note that junior associates learned the craft by doing research and drafting grunt work. If AI absorbs that work, Fitzpatrick concedes firms 'are struggling' to see how future specialists will be trained, though he expects them to adapt. 3. **LexisNexis needed an army of attorneys, not just engineers** Fitzpatrick says he expected to hire data scientists but ended up hiring many practicing lawyers to review AI output by task, such as M&A experience for deal documents. He calls attorney review a 'secret sauce' behind improving results. 4. **The company picks the best AI model per task** A planning agent splits a request into parts and routes each to a suitable model. Fitzpatrick's example uses OpenAI's o3 for deep research and Claude for drafting, so no single model's 'voice' dominates the output. 5. **Fitzpatrick offers no firm limit on what the AI drafts** Pressed on originalism and birthright citizenship, he calls the tool neutral like a brick, cites responsible-AI principles, and says LexisNexis does not interpret law. He could not name any argument or motion the system would refuse to draft. ## Questions this episode answers ### Why can't lawyers just use ChatGPT for legal research? Fitzpatrick says general chatbots are built to give answers that are probably right, which is not enough in court. They can invent cases, work from outdated data, lack attorney-client privacy and can't show their reasoning. ### What does LexisNexis do to stop AI from making up fake cases? Fitzpatrick says answers are grounded in a curated collection of about 160 billion documents and records. Citations link to real cases, and a 'citator agent' checks that each case exists and is still good law. ### Will AI replace junior lawyers? Fitzpatrick says AI augments lawyers rather than replacing them, but he admits it threatens the apprenticeship system. Junior associates once learned by doing research and drafting, and firms haven't yet found a solution. ### Should judges use AI to write their opinions? Fitzpatrick says judges and clerks can use AI to help structure a draft but must review it and take responsibility. He says it should never be 'a bot talking to a bot'. Patel would prefer judges write their own opinions. ### What is the concern about AI and originalism? Originalism says laws mean what they meant when enacted, and some judges want AI to analyse historical word usage. Patel worries courts will defer to an AI they can't hold accountable, and Fitzpatrick says LexisNexis only supplies content, not decisions. ## Facts - **Sean Fitzpatrick** [0:07:22]: "in 2023, uh, when we launched Lexis Plus AI, that's when we became really the an AI powered" - **Sean Fitzpatrick** [0:07:22]: "kind of in in uh 2020, when we launched our Lexis Plus product, uh, we integrated all those things together." - **Sean Fitzpatrick** [0:11:03]: "we used to pay $20, like two years ago for a million tokens, and today, you know, we might pay ten cents for a million tokens, right?" - **Sean Fitzpatrick** [0:10:40]: "each one of them is putting in, you know, uh, close to $100 billion, you know, Microsoft, uh, Amazon, Google." - **Sean Fitzpatrick** [0:13:41]: "system is backed by 160 billion documents and records, you know, that curated collection" - **Sean Fitzpatrick** [0:13:41]: "we call a citator agent, right? So we'll check that case to make sure that it wasn't fabricated by the system, but and that it's actually still good law." - **Nilay Patel** [0:16:24]: "there was just a case where I I believe a court had to rescind an opinion because it had a hallucinated case citation in it." - **Sean Fitzpatrick** [0:23:16]: "some attorney submitted a, um, a document to the court. I think it had like eight citations in it, and seven of them were" - **Sean Fitzpatrick** [0:23:59]: "a couple attorneys got fined 5,000 bucks a piece. And then, you know, some attorneys in federal court, um, down in Alabama got referred to the state bar association for disciplinary action." - **Sean Fitzpatrick** [0:20:54]: "two thirds of attorneys are using AI, uh, in their work, according to to our surveys." - **Sean Fitzpatrick** [0:33:29]: "there's probably eight to 10,000 tasks that could be automated, right?" - **Sean Fitzpatrick** [0:39:22]: "Relics is the parent company, uh, and uh, it's, uh, a publicly traded company. It has four divisions." - **Sean Fitzpatrick** [0:39:22]: "my responsibility is I'm the CEO of our North America, UK, and Ireland businesses" - **Sean Fitzpatrick** [0:43:46]: "But we get eighty of the DNA transferred over in that core model." - **Sean Fitzpatrick** [0:45:09]: "So maybe it uses, um, the O three model from OpenAI, because it's really good at deep research." - **Sean Fitzpatrick** [0:45:09]: "so maybe it uses Claude, uh, three to do that, uh, like the Claude three Opus, which is really good, uh, at drafting." - **Sean Fitzpatrick** [0:47:57]: "our outputs are attorney reviewed, and so that's how we keep getting the, uh, more relevant, um, results." - **Sean Fitzpatrick** [0:49:11]: "it's the fastest growing product that we've ever had." - **Nilay Patel** [0:54:37]: "John Bush is a Trump-appointed judge. He's he cited the emergence of corpus linguistics in the legal field." - **Sean Fitzpatrick** [0:59:22]: "So we always incorporate human oversight, uh, into the, um, development of our products." - **Nilay Patel** [1:00:35]: "now the Trump administration is saying there's nope, actually, that's not what subject to the jurisdiction thereof means." - **Sean Fitzpatrick** [1:01:09]: "These exceptions include foreign diplomats, uh, children of foreign diplomats, children of enemy forces in hostile occupation, children born in foreign public ships" ## Transcript **[0:00:00] Ad read:** Support for the show comes from Comcast Business. With advances in AI, cybersecurity is getting more powerful, but so are attacks. In the past twelve months, Comcast Business detected 24.5 billion phishing attempts designed to exploit user behavior, steal credentials and compromise devices. AI is forcing a shift in how organizations defend themselves, demanding not only faster responses, but continuous intelligence. To survive in the current climate, leaders must build a security first culture to equip their team with the tools they need to stay protected. **[0:00:32] Ad read:** With Comcast Business, connectivity, intelligence and response work as one orchestrated system. Get better equipped to handle today's cyber threat landscape with the 2026 Comcast Business Cybersecurity Threat Report. Read it today at Comcast Business dot com slash cybersecurity. Cyber confidence starts with Comcast Business. Black Friday is around the corner, your biggest chance to grow sales this year. Don't wait. Realize is Taboola's advertising platform for businesses, letting you launch campaigns on top premium websites beyond search and social. **[0:01:09] Ad read:** Add Realize to your media mix and scale this Q4. And as an exclusive offer for podcast listeners, new customers get $700 credit back. Just search Realize Podcast on Google or use the promo code Podcast twenty six. Don't miss out on this limited time offer. Terms apply. Support for the show comes from ServiceNow. AI is moving fast across the enterprise, but without visibility, it's just chaos. Different tools, different models, different teams using AI in completely different ways. **[0:01:45] Ad read:** ServiceNow turns that chaos into control. With the AI Control Tower, you see all your AI across the business in one place. What it's doing, what it's done, and what it's about to do, so you stay in control. To put AI to work for people, visit servicenow dot com. Hm. **[0:02:07] Nilay Patel:** Hello and welcome to Decoder. I'm Nilay Patel, editor in chief of The Verge, and Decoder is my show about big ideas and other problems. Today I'm talking with Sean Fitzpatrick, the CEO of LexisNexis, one of the most important companies in the entire legal system. For years, including when I was in law school, LexisNexis was basically the library. It's where you went to look up case law, do legal research, and find the laws and precedents you would need to be an effective lawyer for your clients. **[0:02:32] Nilay Patel:** There simply isn't a lawyer today who hasn't used a Lexis tool. It's fundamental infrastructure for the legal profession, just like email or a word processor. But enterprise companies with huge databases of proprietary information in 2025 can't resist the siren call of AI, and LexisNexis is no different. You'll hear it right away. I asked Sean to describe LexisNexis to the audience, and the first word he said wasn't law or data, it was AI. The goal is for the LexisNexis AI tool called Prodigy to go beyond simple research and help lawyers draft the actual legal writing they submit to the court in support of their arguments. **[0:03:10] Nilay Patel:** That's a big deal because so far, AI has created just as much chaos and slop in the courts as anywhere else. There is a consistent drumbeat of stories about lawyers getting caught and sanctioned for relying on AI tools that cite hallucinated case law that doesn't exist. And there have even been two court rulings retracted because the judges appeared to use AI tools that hallucinated the names of the plaintiffs and cited facts and quoted cases that didn't exist. **[0:03:36] Nilay Patel:** Sean thinks it's only a matter of time before an attorney somewhere loses their license because of sloppy use of AI. So the big promise LexisNexis is making about Protege is simply accuracy, that everything it produces will be based on the real law, and much more trustworthy than a general purpose AI tool. You'll hear Sean explain how they've built their AI tools and teams so that they can make that promise. LexisNexis has hired many more lawyers to review the work of AI than Sean expected, for example. **[0:04:05] Nilay Patel:** But I also want to understand what Sean thinks tools like Protege will do to the profession of law itself, to the job of being a lawyer. If AI is doing all of the legal research and writing you normally have junior associates doing, how will those junior associates learn the craft? How will we develop new senior associates without a pipeline of junior people in the weeds of the work? And if I'm submitting AI legal writing to a judge who might be using AI to read it, aren't we getting a little close to automating too much of the judicial system? **[0:04:32] Nilay Patel:** These are big questions, and they are coming to a head real fast in the legal industry. I also pressed Sean pretty hard on how judges, particularly conservative judges, are using computers and technology in service of a judicial theory called originalism, which states that laws can only mean what they meant at the time they were enacted. We've run some stories here at The Verge about judges letting automated linguistic systems try and understand the originalist intent of various laws to reach the preferred outcomes. **[0:04:59] Nilay Patel:** And AI is only accelerating that trend, especially now in an era where literally every part of the Constitution appears to be up for grabs before an incredibly partisan Supreme Court. So I asked Sean to demo Protege doing some legal research for me on a question that appears to be settled, but is newly up for debate in the Trump administration. The question of whether the fourteenth Amendment guarantees birthright citizenship to everyone born in the United States. To his credit, Sean was game. **[0:05:25] Nilay Patel:** He did it. But you can see how taking LexisNexis from a company that provides simple research tools to one that provides actual legal reasoning with AI will have big implications across the board. This conversation is weedsy, but it's important and it touches on so many things that we've talked about here at Decoder. Okay, LexisNexis CEO Sean Fitzpatrick. Here we go. Sean Fitzpatrick, you are the CEO of LexisNexis. Welcome to Decoder. Thank you. Great to be here. **[0:06:05] Nilay Patel:** Thank you for for joining me. I was just saying this is my first interview back from parental leave. So, uh, apologies if I'm rusty to the audience, but apologies to you, uh, if I'm just like totally loopy. But I I I Congratulations. I'm Yeah, I'm very excited to talk to you. I feel like the legal profession in America, you know, I I'm very much a failed lawyer. My wife is a lawyer. There's a lot of lawyers on the Verge team. **[0:06:27] Nilay Patel:** The legal profession in America is just at a moment of absolute change, a lot of chaos, actually, and an enormous amount of uncertainty. And LexisNexis, if the audience doesn't know, tends to sit at the heart of what lawyers do all day. But most lawyers are using LexisNexis all day long, every minute of every day. And what that product is and what it can do and how it can help lawyers do their job connects to a lot of themes that I think we see both in in the legal profession and then with technology and AI generally. **[0:06:57] Nilay Patel:** So just start at the start. What is LexisNexis? How do you explain it to the layperson? **[0:07:02] Sean Fitzpatrick:** LexisNexis is uh an AI powered uh provider of information and analytics and, um, drafting solutions, uh, for lawyers, uh, that work in law firms and corporations and government entities. **[0:07:16] Nilay Patel:** That's a new conception of LexisNexis. When I was in law school in the early 2,000, it was just the thing I searched to find case law. **[0:07:22] Sean Fitzpatrick:** Yes, it's we've, we've transformed over time that we were kind of just that research provider. Uh, you know, over time, we've acquired more businesses, we've integrated those businesses, and kind of in in uh 2020, when we launched our Lexis Plus product, uh, we integrated all those things together. Uh, and so we became an integrated ecosystem of solutions. And then in 2023, uh, when we launched Lexis Plus AI, that's when we became really the an AI powered, um uh, provider of information analytics, decision tools, and and drafting solutions. **[0:07:56] Sean Fitzpatrick:** And, you know, the capabilities of AI have really allowed us to do things, do more things, uh, than what we've traditionally done in the past. **[0:08:02] Nilay Patel:** That jump from being the, you know, sort of gold standard database of legal opinions and reasonings and case notes and all all that to we're gonna do the work for you or we're gonna help you do the work. That's a that's a big one. That's a cultural jump. Obviously, there's some acquisitions along the way you can talk about that helps you make that jump. What drove you to make that jump? To say, actually, the lawyers need help drafting the proposed opinions they might give to a judge. **[0:08:26] Nilay Patel:** What made you say, Okay, we've got to step into actually doing the work? **[0:08:29] Sean Fitzpatrick:** Yeah, I mean, I think it's been a natural evolution, and as as technology has evolved, um, it's opened up new, uh, avenues of things that we can do. What what we do is we tend to, um, take the latest technology and we introduce that to our customers, and we spend time talking to them about how they think that that technology can be best applied in the um, in the uh, legal environment. Uh, and then based on the ideas that they come up with, we, um, transition uh, or or uh, translate that into products and build products that, uh, that, uh, resolve those those, um, opportunities or address those opportunities. **[0:09:05] Nilay Patel:** Let me ask you a pretty philosophical question. It's one that I I struggle with all the time. It's one that I talk to our audience about all the time. Our audience is pretty technically focused. I think most people who encounter the legal system think it's pretty deterministic. People Our audience is pretty technically focused. They're used to computers. Computers are until recently pretty deterministic, right? You put in some inputs, you get some outputs. Most people who encounter the legal system think it's pretty deterministic. **[0:09:30] Nilay Patel:** You put in some inputs and you get some predictable outputs. And what I'm always saying is, well, that's not how it works at all, right? You show up to court, the judge is in a bad mood, you have no idea what's gonna happen. You show up to you're a big company and you have an antitrust appeal, and you show up to the three judge appellate appellate review board, and you have no idea what's gonna happen. Anything, like literally anything could happen anytime. **[0:09:52] Nilay Patel:** And the judicial system is fundamentally not deterministic. And trying to think about it like a computer, even though it's structured like a computer, can get you in all kinds of trouble. Uh, maybe the best example of this is, uh, people on Facebook putting the words no copyright intended on the bottom of movies or part. Like they think they can issue these magic words and the legal system is solved, and they just can't. AI, like, is that problem in a nutshell, right? **[0:10:17] Nilay Patel:** We're gonna take a a computer, we're gonna make it better at natural language. We're gonna fundamentally make the computer not deterministic, right? You can't really predict what an AI is gonna do. And then we're gonna apply that to the fundamentally not deterministic, the very human nature of the court system. Somewhere in there is a big philosophical problem about applying computers to the justice system. How do you think about that? **[0:10:40] Sean Fitzpatrick:** uh, first of all, you these you have these massive investments that are happening with the foundational models, right? Uh, these hyperscalers, you know, each one of them is putting in, you know, uh, close to $100 billion, you know, Microsoft, uh, Amazon, Google. And so these models, they just, you know, continue to get better and better and better over time. And that's at the foundational model level. We don't really operate at the foundational model level. **[0:11:03] Sean Fitzpatrick:** We build applications, uh, that utilize these foundational models. And and at that level, what we see is prices are dropping, right? Uh, so we used to pay $20, like two years ago for a million tokens, and today, you know, we might pay ten cents for a million tokens, right? So that allows us to do things at speed and at scale that we've we've never been been able to do before. And and if you look at the the law, you know, there are a lot of things, um, about the law that make these models attractive. **[0:11:32] Sean Fitzpatrick:** So most of the law is language based, right? And these models are really, uh, really great with with language process uh uh problems. Um, the law is precedent based, right? And so, **[0:11:42] Nilay Patel:** Well, uh, we we'll get we'll come to that. I'm not sure that that that's under, uh, that's up for grabs, basically. We'll, we'll, we'll come back to that. **[0:11:49] Sean Fitzpatrick:** I'll grant you that, yeah. Uh, and then you look at the activities that lawyers do, you know, they, um, draft documents, they do, uh, research, they summarize things. The models are all, like, really, really good at at these these types of things. Uh, and so you kinda have this like perfect storm of this technology, um, and the things that legal do, like lawyers do, c kind of coming together. And yet, when people try to use these models, these consumer grade models, um, there are all kinds of problems with them, right? **[0:12:19] Sean Fitzpatrick:** Uh, you can't just put it like you said, it's not deterministic. You can't just put information into a computer and get an answer out. If that were the case, we wouldn't need a court system, right? Uh, but, you know, what what we see happening, uh, is, um, you know, w like with these models, like, th they're just not built for the legal system. So, uh, you can't go into court and say, I found this on the internet, right? **[0:12:43] Sean Fitzpatrick:** You, you have to have authoritative content. Um, the cutoff date for, uh, GPT, uh, four o was 2023, I believe, right? You need to have information that's that's constantly updated. Um, you know, your audience probably doesn't know, you probably know this because you're a lawyer, but, uh, there's the citator, right? You, you know, which traditionally has said this is good law or it's not good law, right? It's been overturned. Now it'll tell you if it's a law at all, right? **[0:13:12] Sean Fitzpatrick:** Uh, or if the the you know, some system just just made it up. Because, you know, these, uh, systems, they're probabilistic, right? They wanna, um, put to any together an answer that's probably right. Well, that's that's not the standard that we have in in legal. You can't go in, uh, with something that's that's probably right. Um, and so you have this whole list of of of issues these models don't address. And so what what we've tried to do is address those with a courtroom grade solution. **[0:13:41] Sean Fitzpatrick:** So our our system is backed by 160 billion documents and records, you know, that curated collection, uh, that we have. That's our our grounding data, right? So you can go into court and not say I found this on the internet, but you could say you can refer to a specific case, right? Um, we also have a, um, we call a citator agent, right? So we'll check that case to make sure that it wasn't fabricated by the system, but and that it's actually still good law. **[0:14:06] Sean Fitzpatrick:** And you can also look at the case law summary. You can so you know what the case is about. You can look at the headnotes, so you can see the particular points of law that were addressed in that case. And you, again, you can see if it's if it's still a valid case. Um, privacy is another issue, right? Uh, you know, there's this special relationship that exists between, um, attorneys and their clients, and that attorney-client privilege. **[0:14:30] Sean Fitzpatrick:** Um, you know, you, there's some privacy requirements, uh, that you need to have in order to maintain that. Uh, you know, if you're using one of these just consumer grade models, you you don't have that level of of, uh, of privacy and, um, security that you need. Um, transparency is another issue. So, uh, you know, you put this question in, you get an answer back. Well, well, based on what? Like, what was the logic that the the system used? **[0:14:56] Sean Fitzpatrick:** Um, so our again our system, we open up the black box and you can see the logic that's being applied. And we give the attorneys the ability to go in and actually change that, right? If it if this model is getting something wrong, um, the attorney has the opportunity to change it so that they get the the outcome that they're they're really driving for. But as you said, like it's not the law is not deterministic, right? **[0:15:17] Sean Fitzpatrick:** There are lots of different factors that go into this, but you need to have a system that's legally driven, you know, that's purpose built for legal situations, uh, in order to, you know, to really operate in a in a courtroom type environment. **[0:15:31] Nilay Patel:** Yeah, there's two things there I really wanna push on. One, again, I I was not a good lawyer. I I don't I don't wanna ever pretend on the show or to you or to anyone that I w- was any good at this. But you know, the thing you learn in law school is a particular way of thinking, which is a pretty rigorous, structured way of approaching a problem and then going to find the relevant cases and precedents, and then trying to fashion some solution based on that. **[0:15:52] Nilay Patel:** That feels like, uh, we're just moving words around, but it's actually a way of thinking, right? And the before AI showed up, the we're gonna use a word processor and we're gonna think in a certain way. They were mashed together, and now we're pulling them apart, where we're saying that the computer can move the words around and generate you some thinking. So that's one thing I wanna I wanna push on. I'm very curious about that because it feels like the the lawyering part of being a lawyer is being subsumed into a system, and that might change how we lawyer. **[0:16:24] Nilay Patel:** Very curious about that. The other part is, well, is anyone gonna look at the work being done? Cause we're already seeing lawyers get sanctioned for filing briefs with hallucinated case citations in them. Uh, there was just a case where I I believe a court had to rescind an opinion because it had a hallucinated case citation in it. This is bad, right? Like, this is just straightforwardly a threat to the system and how we might think about lawyers and judges and courts. **[0:16:52] Nilay Patel:** And it it's not clear to me that anyone's gonna use the tools as rigorously as you want. So on the one hand, there's the we've made the thinking easier, and on the other hand, it's, oh boy, everyone's gonna get really lazy. And they're both kind of in your answer, right? They're they're both kind of like, what are we? We're making it easier to look at this stuff, we're making it faster to do the research. And I'm just wondering where you think the thinking comes in. **[0:17:13] Sean Fitzpatrick:** yeah, I think the these models, they don't replace lawyers, right? I think they help the lawyer and they augment what the lawyer does. So if you think about an activity that a lawyer might do, like, let's say that they were, um, preparing for a deposition. So they need to come up with a list of questions that they're they're gonna ask the individual that's being deposed. Um, you know, you can take the, um, facts, uh, around that particular case, and you can load them into, um, a vault, and then you can point the system to that vault and say, based on the facts of this particular matter, develop a a list of of deposition questions. **[0:17:49] Sean Fitzpatrick:** Um, and then, you know, that's something that a lawyer would have done on their own, right? Uh, in the past, so they may have, you know, referred to a list of questions that they had previously or something. **[0:17:58] Nilay Patel:** Well, actually, can can I just grab on that example? May maybe a lawyer would have done that, but m more often, a lawyer would have told a bunch of junior associates to sit in the basement and do that, right? And that thing was how those junior associates learned how to do their job. And that's what I mean. Like, we're we're sort of farming out the thinking and some people might never actually do that thinking. And that might change the profession down the line in in really substantive ways. **[0:18:25] Sean Fitzpatrick:** right, yeah. And it is a an apprentice, uh, system, right? And so if you start to take some of the layers out of the bottom, you know, how how does everyone s, you know, skip the bottom layer and still make it to the second layer with the same level of, um, uh, capabilities and skills? I think I think that's a real challenge. Um, you know, I think the systems are allowing lawyers to not necessarily, uh, have the associate do that work, but now they can say, you know, generate me 300 questions or 700 questions. **[0:18:57] Sean Fitzpatrick:** You know, it doesn't it doesn't take that long to go through 700 questions, and the models never get tired. So, and and what our experience is, is they'll they'll go through that list of questions and they'll say, first question, yep, that's a good question. I would have thought of that, you know? So the system made it a little bit faster, right? But it didn't really help them any. Uh, second question, same thing, third question, same thing, fourth question doesn't even make any sense. **[0:19:22] Sean Fitzpatrick:** Scratch it off the list, right? Fifth question, oh, that's that's interesting. I wouldn't have thought to ask that, but that's that's really, um, you know, something that's probably important, so I'm gonna add that to my list. So there's an efficiency component to it, but I think there's also a better, better outcome, um, component to it. In terms of the apprenticeship piece of it, uh, yeah, I think people are struggling right now to figure out how that's going to, um, impact the apprenticeship model. **[0:19:50] Sean Fitzpatrick:** And if you don't have people, someone was describing to me that they had worked on a a situation where they they were looking at, um, securitized assets, uh, and collecting debt, you know, on securitized assets. And when they were associate, they did this project for a company that had states, you know, fifty states, uh, worth of coverage. And so they became the expert in the firm on asset securitization in all fifty states. And for four or five years, anytime somebody had a question, they came to to that individual. **[0:20:18] Sean Fitzpatrick:** It was a great way to make a career. Now that all the system can do all that that information for you. And so his question was like, like, how, uh, how is that ever gonna happen now in in this new world? I think I think firms are gonna struggle with that, but I also think they're going to figure it out. Uh, you know, we tend to get, you know, the some of the smartest and and brightest people, uh, going into the legal profession. **[0:20:42] Sean Fitzpatrick:** Um, and so far, they seem to have figured out every challenge that's, uh, that's faced the industry in the past. I think they'll they'll figure this one out, um, as well. **[0:20:51] Nilay Patel:** What are some solutions that you've seen as people try to figure this out? **[0:20:54] Sean Fitzpatrick:** Well, I don't I don't know that, um, necessarily, uh, folks have come up with a lot of solutions around the, um, apprenticeship model. I think that that what we're seeing for sure is people are embracing AI. Uh, you know, it's here, it's in the courtroom, it's in the law firm. Um, you know, two thirds of attorneys are using AI, uh, in their work, according to to our surveys. Uh, they're already using that, and and our survey is probably a little outdated. **[0:21:19] Sean Fitzpatrick:** I'd say the number's probably higher. You know, I I don't know about you, but I use, uh, AI every day, practically, uh, is in my personal life and in my work life now. Uh, and I think the legal profession is, you know, perfectly suited for it. So I think it's, uh, it's only going to expand. **[0:21:37] Nilay Patel:** When you see the lawyers getting sanctioned and the courts having to rescind opinions, is your solution well, you should just use Lexus and that won't happen to you? Or do you think that's a the symptom of something else? Cause that everyone's just using AI? I get it. I I think probably the biggest split in our audience right now is the data that says everyone's using the stuff all the time, and the hostility our audience expresses to us about the tools, their quality. **[0:22:02] Nilay Patel:** The fact that a lot of that usage is driven by the by the, you know, big companies just putting in front of you, whether you want it or not, right? There there's something happening there where to justify these enormous investments, the tools are showing up whether the consumers are asking for them or not. And then we're pointing at, well, everyone's using the tools. And what I hear from our audience is well, I I I can't turn off the AI overview. **[0:22:23] Nilay Patel:** Of course, I'm using the tool because it's just in front of me all the time. I I can't make Microsoft Office stop telling me that it's gonna help me do stuff. It's just in front of me all the time. So for you, when you see the errors being made in the legal system today, right? The the lawyer's getting sanctioned, the lazy use of AI, the lack of apprenticeship, which is gonna impact the entire next generation of lawyers and how rigorous they are. **[0:22:45] Nilay Patel:** How do you make your product address that, or are you just not thinking about that right now? **[0:22:49] Sean Fitzpatrick:** No, we're definitely we're definitely thinking about it, and, uh, we've we've incorporated things into our product. You know, I I think it's a small percentage of attorneys, uh, you know, these things always make the headlines whenever one one happens, but I think it's a small percentage of attorneys that are that are causing these problems. Most of them, you know, it's it's never been the standard that you just take something and bring it into court, right? Like you've always had the responsibility as a lawyer to check that material and make sure that it's valid before going into court. **[0:23:16] Sean Fitzpatrick:** Uh, and some individuals aren't doing that. We certainly saw that, like in the Burke case, um, where, you know, they, you know, some attorney submitted a, um, a document to the court. I think it had like eight citations in it, and seven of them were, um, just, um, completely, uh, **[0:23:32] Nilay Patel:** But that was inevitable, right? Like, uh, the day ChatGPT showed up, like half of the legal pundits I know are like, it's this is inevitable, like this outcome will happen. And then it it happened, like it was. There wasn't even like a stutter step. It just happened immediately. And that's what I'm I'm trying to push on is is the solution just LexisNexis has a tool that's better and you should pay for it? Or is the solution that as we take the rigor away from the younger associates, the profession is going to have to build some new guardrails? **[0:23:59] Sean Fitzpatrick:** yeah, well, like, you can never stop an attorney from, um, just taking it into court and not doing their, uh, you know, the the proper work, right? I think that's going to continue to happen. I think somebody's going to lose their license over this at some point, right? And we're seeing the sanctions start to to ratchet up. So, you know, it was, you know, a couple attorneys got fined 5,000 bucks a piece. And then, you know, some attorneys in federal court, um, down in Alabama got referred to the state bar association for disciplinary action. **[0:24:27] Sean Fitzpatrick:** You know, I think the stakes are are increasing and increasing. With with our system, what we do, uh, is if we have a citation, uh, we will provide a link to that citation. So you can click on it and you can see it in our system. And there's no fabricated cases within our system. You know, we have a a collection mechanism that ensures that that every case in there is a valid case. It's Shepardized, it has said note headnotes and and, um, uh, you know, different tools that the the lawyers can use. **[0:24:55] Sean Fitzpatrick:** So we're making it, like, really easy if you use our system to check to make sure that the, um, citations that you're, um, you know, that you're bringing into court. Not not only are they valid, and there's, you know, there's still good law, but they're they're in the right format, right? Like format's important. Um, uh, and so, you know, we we check for all these things and make it really easy for the lawyer to to do the work that they need to do right. **[0:25:19] Sean Fitzpatrick:** They need to make sure that case is on point, that that case is still valid. **[0:25:25] Nilay Patel:** We have to pause here for a quick break. We'll be right back. **[0:25:36] Ad read:** Support for the show comes from Upwork. Every business owner hits a point where they need more expertise than they can handle alone. But another full time hire isn't always the answer. That's where Upwork comes in. With Upwork, you can hire highly skilled freelancers across a variety of specialties, from software development and AI implementation to marketing, design, and business operations. You can browse profiles, review past work, and get help scoping the role. That way, you can hire with confidence and get started quickly. **[0:26:07] Ad read:** And with Business Plus, you can access the top 1% of talent on the platform. Thousands of growing businesses already trust Upwork to hire highly skilled freelance professionals. It's free to sign up, and posting a job is simple. Visit Upwork dot com right now and post your job for free. That's Upwork dot com to connect with top talent ready to help your business grow. Spelled U P W O R K dot com. Upwork dot com. **[0:26:37] Speaker 6:** This is advertiser content from Comcast Business. **[0:26:37] Ad read:** Hi everybody, welcome back to another edition of Decoder Sessions. This segment features Andrew Melisak, the Verge's general manager, and Ivan Shefrin, the executive director for managed security services at Comcast Business. I hope you enjoy this conversation. Let's talk about artificial intelligence. I want to understand how has AI, um, you know, changed the cybersecurity landscape for you and your team at the attack level? **[0:27:09] Interviewee:** The reality is it has had a dramatic impact on everything we do. The bad guys, uh, leverage, uh, leverage AI to for malicious and bad purposes. So the only way to, uh, uh, fight fire is with fire. And we so we now use AI just as the attackers are using it to break in more quickly, we have to pivot and use it to defend more quickly at scale. **[0:27:35] Ad read:** Comcast Business specifically detected 25.4 billion phishing attempts and 21.9 billion drive by compromise events over one year. Uh, talk to me a little bit about that data. **[0:27:49] Interviewee:** Phishing is so popular because it's so easy to do, and AI has accelerated that. Every business should have, uh, software in place to detect phishing attacks, uh, to train their users, and also that now extends to collaboration software, not just email, such as Zoom, such as Slack, such as Teams, um, because that's becoming increasingly prevalent. Everybody expects to get breached now, and that is not the attitude we use at Comcast. We we are take protecting our customer data extremely seriously. **[0:28:33] Ad read:** Hm. Support for the show comes from Granola. Have you ever left a meeting thinking that was productive, and then ten minutes later, when it's time to take next steps, you're at a loss. Or worse, you've been heads down the entire meeting, furiously typing notes instead of actually listening, and now your brain is fried. Meetings are a mess. Granola fixes that. Granola is an AI power notepad that makes your meeting notes useful. It capture what happens in your meetings and turns it into clean, structured notes with decisions and action items pulled out and made easy to find. **[0:29:05] Ad read:** And the best part, Granola works through your device's audio, which means it integrates seamlessly into the video conferencing tools you already use. No setup, no awkward bots. It's just your normal meeting with superpowers. You get to actually listen instead of frantically typing every word, and still walk away knowing exactly what was decided, who's doing what, and what comes next. Once you try it on your first meeting, it's hard to go without. You could try it totally free. **[0:29:32] Ad read:** Just head to granola dot ai slash decoder. That's granola dot ai slash decoder to get your time back. Go to granola dot ai slash decoder and try it on your next call. Hm. **[0:29:50] Nilay Patel:** Welcome back. Right before the break, I was talking to LexisNexis CEO Sean Fitzpatrick about how lawyers really shouldn't bring the totally fabricated research they do on ChatGPT into court. And that led to the obvious next question. What other parts of the process does Sean see AI and LexisNexis taking over? One of many reasons I was a horrible lawyer was that moment when you get your first law firm job and you realize, oh, my boss just has a library of their favorite motions on file. **[0:30:22] Nilay Patel:** And they're gonna just pull, they're just gonna pull from the card catalog, and they're gonna change, you know, some names and dates, and they're gonna file this motion. And the judge will recognize the motion and the attorney, and this is all just like a weird formality to get through the next stage of the process, right? And maybe we'll never get to the substantive part of the case cause we're just gonna settle it, but we need to file this motion we add in the library. **[0:30:42] Nilay Patel:** This really, like, truly was demoralizing. I was like, I'm just doing paperwork. Like, there's nothing about this that that is real, like in a in a way. I'm probably describing what every first year associate goes through until the check hits, and I I it just didn't work for me. How close are you to having the Lexus AI product? Just do that thing. Just recognize the moment and say we have the banked motion and we're just gonna file it to a system. **[0:31:11] Sean Fitzpatrick:** well, but we're connected, uh, you know, we can connect into a DMS, a document management system that has an attorney's prior motions. We have our vault vault capability so they can load their emotions up so they can still use the motions that they that they've already developed, right? And that's that's a perfectly fine way to do things because **[0:31:32] Nilay Patel:** Well, I'm saying from from scratch, from, you know, if **[0:31:34] Sean Fitzpatrick:** right, but yeah, we have the ability to do it from scratch too. Uh, and so, uh, but a lot of times they don't want to do it from scratch because they've reviewed every single word in that motion and they know that it's good. And if they do it from scratch, then they have to review every single word, right? Uh, but if they want to do it from scratch, uh, we can do that for them, uh, today. **[0:31:54] Sean Fitzpatrick:** And we can use their prior work product if they want to use that as the, um, grounding content, uh, to to create a new motion, or we can use our authoritative material. They can choose, um, what the source is, what the grounding content is. **[0:32:09] Nilay Patel:** I guess I'm asking, what's the level of automation there, right? So you're an attorney, you've got a document management system, you've got a new client. You need to file some standardized motion that you always file for whatever thing you need to do, a continuance. At what point does Lexus say, I'm watching this case, I'm gonna file this for you. I'm just I'm just gonna hit the buttons for you. Don't worry about it. In a way that a great legal assistant might do that. **[0:32:31] Sean Fitzpatrick:** Yeah, we're always going to give the, uh, attorney, um, the opportunity, like, we we don't want to just be doing things on their, uh, behalf, uh, in an unsupervised way, right? So we're gonna give them the opportunity. We we, you know, we could get to the point where we say, Hey, look, uh, it looks like you need a continuance. Uh, here's a draft of a continuance. Push here, and, you know, it will automatically file it. We're not to that point today, but if you need a continuance, we could draft it for you, right? **[0:32:59] Sean Fitzpatrick:** Um, and if you think about like, like our our vision is that every attorney is going to have their own personalized AI assistant, and it's going to understand their practice area, it's going to understand their jurisdiction, it's going to have access to their prior work product, it's going to have access to, and systems are only as good as the content behind them, right? So it's going to have access to our 160 billion, um, documents and and records, um, and it's going to be able to automate tasks that they do today, right? **[0:33:29] Sean Fitzpatrick:** And, uh, if you think about all the different types of attorneys and all the different tasks that they perform, you know, there's probably eight to 10,000 tasks that could be automated, right? And so we're we're working with our customers to understand, you know, what are the most important tasks, and we're working with them to to automate those tasks today. So we have, you know, the largest and most robust backlog of, uh, pro projects, uh, that we've ever had in the history of our company. **[0:33:58] Sean Fitzpatrick:** Because there's so many of these things that, uh, that still, um, are to be automated. But, you know, we're working, uh, we're working with our customers to do that. If they tell us that, hey, what we really want is for you to automatically file this, or for you to, you know, provide me with a, um, an alert, uh, that says, Hey, this this deadline is coming up and, uh, you need to file this. Here's a draft. **[0:34:21] Sean Fitzpatrick:** Do you want to file it? If they if they ask us for that, I'm sure we can develop it. We're we're not we're not at that point today. We're, but we are in in the drafting phase. And then that that vision, that's not like a five year vision or a three year vision. That's available today. That's Protege, right? That's what Protege does today. But it's it, you know, there there are tasks that it can do, but we haven't finished that, you know, that massive backlog yet. **[0:34:43] Nilay Patel:** Yeah, that's what I was gonna. Yeah, I mean, if you look at the suite of other CEOs who've been on decoder, you know, they're gonna tell you it's just a you just integrate our computer vision system and we'll use ECF for you to file this motion. And it's it's all they're all they're all they'll all be very happy to sell you that product, I'm sure. But the reason I'm asking it this way is when I get the consumer AI CEOs, uh, they love to tell me that they're gonna write my emails for me with AI. **[0:35:06] Nilay Patel:** And then the next sentence they say is, and then we'll sort your inbox with AI. And at some point, the robots are just writing emails to each other, and I'm reading summaries, and I something very important has been lost in that chain. And this is one of the funniest outcomes of AI is suddenly my iPhone is just summarizing emails and generating emails for other iPhones to summarize, and I have no idea what's going on. That's bad in the legal context, right? **[0:35:31] Nilay Patel:** We're we're automating the the generation of the documents to make the case for our clients. And on the other side, the judges and clerks might be using these same tools to ingest them, summarize, understand the arguments, and write the opinions that are the outcomes. Where do you see, like, culturally, I think it's important for you to have a point of view on where that should stop. Because otherwise, we are just gonna have a fully automated justice system of LLMs talking to each other, maybe with some guardrails that other people don't have. **[0:36:03] Nilay Patel:** But we've taken an enormous amount of humans out of the loop. **[0:36:06] Sean Fitzpatrick:** yeah, no, I think you have to have the human in the loop, uh, that it's an important part of the process. You know, if if someone says, Hey, can you meet at nine o'clock? Uh, and your system opens up your calendar and says you're available to meet, and you've got that person on your high priority list and it sets up the meeting. I mean, I could see bots going back and forth to do those kind of things. **[0:36:25] Sean Fitzpatrick:** When you're talking about, like, substantive, um, legal matters, the stakes are too high. You know, you're you're talking about, you know, a disabled veteran, you know, getting their benefits or not getting their benefits. You know, you're talking about a victim of a natural disaster, you know, getting insurance proceeds or not getting in insurance proceeds. You're talking about a single mother, you know, being, you know, uh, getting, uh, welfare benefits or not getting welfare benefits. These are all legal matters, and they're they really have a huge impact on people's lives. **[0:36:54] Sean Fitzpatrick:** The stakes are, I think, are way too high for bots to be going to going back and forth and sharing information. And and they **[0:36:59] Nilay Patel:** Do you think that clerks and judges should be using AI the same way the lawyers should be? Cause, uh, that's where I would draw the line is, I don't I think the clerks should be made to read everything as humans and interpret everything as humans. The judges should be made to write everything as humans. But it doesn't seem like that line has been formalized anywhere. **[0:37:16] Sean Fitzpatrick:** I I don't think that they should, like, I don't think a judge should write every line. Um, I think that they could use AI, you know, it's great when you put concepts in at being able to put the words around that concept and put, you know, uh, structure them in an an orderly way. Um, so I think that there's a, um, there is a component of the work that could be done by AI. Um, but I don't think it should be a bot talking to a bot. **[0:37:44] Sean Fitzpatrick:** I don't think it should be fully outsourced to AI. I think that you've got a responsibility, you know, as a judge, as a law clerk, as a lawyer, uh, to review that document and make sure this is actually saying what you intended to say. Um, and I think most attorneys are are using it that way, right? It will create a great draft, maybe an eighty draft, right? Uh, which allows you to do twenty of the work, but that twenty of the work is like the deep analytical thought work, the things you actually went to law school to do, as opposed to kind of what you were describing earlier, right? **[0:38:16] Sean Fitzpatrick:** Um, and I think it's going to allow lawyers to do more of that type of work. **[0:38:20] Nilay Patel:** I'm curious to see how different jurisdictions and and circuits approach the question of what should the judges be doing and what should the clerks be doing. Cause I I sense that that pressure is going to express itself in different ways across the field. **[0:38:33] Sean Fitzpatrick:** I think it is. Yeah, I mean, I think, yeah, I mean, jo judges are are becoming forensic auditors, right? And they're reviewing this, uh, information, looking for, uh, fake cases, right? That's that we don't want them doing that. That should not be their job, right? I think things do do need to change in some of these areas. **[0:38:48] Nilay Patel:** I'm not 100% sure. Yeah. Go go ahead. Using AI to catch AI is another theme that comes up on Decoder all the time. I will say it is my first interview back after twelve weeks and I've utterly forgotten to ask you the decoder questions. So let me do that and then I want to zoom out a little bit farther. Uh, these are my own questions. You can tell I'm a little rusty. Uh, LexisNexis, I'm looking at the leadership structure. **[0:39:07] Nilay Patel:** It's very complicated, right? There's a CEO who's not you, Mike Walsh, but then you're the CEO of the US and the UK. There's a bunch of other VPs everywhere. You've got a parent company called Relx. Explain how LexisNexis is structured and and how your part fits into it. **[0:39:22] Sean Fitzpatrick:** yeah. So, uh, Relics is the parent company, uh, and uh, it's, uh, a publicly traded company. It has four divisions. Legal and professional is one of those divisions, and the CEO of Legal and Professional is Mike Walsh, and I report to Mike. Uh, and my responsibility is I'm the CEO of our North America, UK, and Ireland businesses. Uh, and so, uh, you know, the way that we're organized, um, we're it it's a matrix, right? So we go to market, uh, based on customer segments. **[0:39:55] Sean Fitzpatrick:** So we have a large law business, a small law business, a corporate legal business, a federal government business, a state and local government business, a Canadian business, a UK business. Um, and then we have functional groups that support that. So we have product management, uh, and then they're responsible for our our product development roadmap and the product strategy. And then we have an engineering team, and they take the direction from product management, but they actually build the products. **[0:40:21] Sean Fitzpatrick:** And then we have functional groups that support that finance, HR, legal, um, global global operations that does things like collects collects content for us. So it's, uh, from the inside out, once you get used to it, it's not that complicated of a structure, and it's really well integrated and and, um, seamlessly integrated together, uh, which allows us to operate r really, um, uh, quickly. Uh, you know, we can get things done quickly, and I think in a in a an efficient way. **[0:40:53] Sean Fitzpatrick:** And I I would say it's it's all customer, you know, the whole process is customer driven. **[0:40:57] Nilay Patel:** So I'm really interested in in the structure, in particular, the fact that you have the UK, Ireland and North America. Well, it the, uh, you know, I'm fascinated by corporate structures. And one of the things that strikes me at this is you are not in control of the taxonomy of your product, right? The the governments of those countries are in control of the taxonomy of their legal systems. The English legal system and the American legal system have commonalities, but wildly different structures, right? **[0:41:25] Nilay Patel:** The Canadian legal system and the United States legal system, wildly different structures, Canada actually has more in common with the UK, given their shared history. How do you think about that? Is are those different teams? Do they have different database structures? How does all that work? **[0:41:38] Sean Fitzpatrick:** yeah, um, we do have different teams and we do have different database structures, but we're actually trying to consolidate to the extent we can. Because, you know, to the extent that we have things that are similar, we shouldn't have them marked up in different ways in in different databases. Getting them marked up in a consistent way will allow us to what we call extreme reuse, but to basically use our, you know, that same technology that we developed in multiple jur uh jurisdictions with limited changes to that system. **[0:42:10] Sean Fitzpatrick:** And what that allows us to do is really focus on that core system, um, and roll it out quickly. And so the everyone across the world, uh, gets the benefit of all those changes. But, you know, you have, you know, um, civil law in in some jurisdictions and, um, you know, uh, common law in others, and uh, you have all the the laws are structured in different ways. And, um, and so you do have things that that make that, um, more challenging. **[0:42:36] Sean Fitzpatrick:** Uh, but that's that's the general idea, um, behind what we're trying to do there. **[0:42:40] Nilay Patel:** Do you think you can you apply the same sort of AI systems to these different legal systems in the same way, or are you actively localizing them in different ways? **[0:42:49] Sean Fitzpatrick:** I, I would say that we actively localize them, but we try to minimize the amount of work that we do to localize them, uh, because a lot of it can be done in a a similar way. **[0:42:59] Nilay Patel:** There's a lot of concern generally about American legal precedence, sort of traveling across the ocean, particularly in the UK. You can see, like, American culture war gets exported a lot, that shows up in a lot of different ways. Do you think your tool will make that better or worse? Right? Cause if you're not pulling them apart, you're actually trying to minimize the differences, you might see repeat arguments or repeat structures just based on the way the AI works. **[0:43:23] Sean Fitzpatrick:** yeah, I mean, each one is bet is based on the content of the individual jurisdiction, so we don't mix the the content. Um, but we do, you know, we do try to utilize the same technology. Uh, so, for example, um, search relevance technology to find the most, um, you know, the case that's, uh, most closely associated with the matter that someone is working on. Uh, you know, we can take that, we can build it like for the US. **[0:43:46] Sean Fitzpatrick:** market, for example, or the UK. market, and then we can move it to another market, and it will work pretty good, right? It will work pretty well. And then we need to do some modifications to make it work really well for that particular particular jurisdiction. But we get eighty of the DNA transferred over in that core model. **[0:44:02] Nilay Patel:** I was talking to Mike Krieger, who's the chief product officer of Anthropic, and it just a totally different conversation on a different thing. But he said this thing to me, which has stuck in my mind. He said, I recognize Claude, I can see Claude's writing, and I'm like, and he said, That's my boy, which is cute. Does your AI have a personality? Can it recognize its writing in all these different jurisdictions? **[0:44:25] Sean Fitzpatrick:** Uh, you know, we use, uh, a multi model approach, um, and so, uh, I think that it's probably a little less, uh, clear, like, which particular, uh, model, um, drove something. So, you know, like, if someone puts in a request, and of course, you know, with agentic AI, things have really changed, right? I think that probably was true, maybe, you know, a year and a half ago, but now with agentic AI, when someone, um, puts in a query, like, let's say they wanted to, um, draft a a document, maybe a client sent in a request, and she's interested in a premises liability issue, uh, you know, around, like, duty to inform a trespasser on, uh, about a dangerous condition on **[0:45:09] Sean Fitzpatrick:** a piece of land, for example, right? Um, the the, uh, the query will go into a, uh, an agent, a planning agent, uh, who will then allocate that, um, query out to other agents. So, you know, it needs to do some deep research. So maybe it uses, um, the O three model from OpenAI, because it's really good at deep research. And, you know, at the end, it needs to, uh, draft, uh, a document, uh, so maybe it uses Claude, uh, three to do that, uh, like the Claude three Opus, which is really good, uh, at drafting. **[0:45:49] Sean Fitzpatrick:** And so what we're at model agnostic, we'll use whatever model is best at a particular task. And so the result that you get back is actually work that's potentially done by multiple different models, which I think probably makes it a little bit harder just to see, like, oh, yeah, I know that was drafted by, um, OpenAI. **[0:46:05] Nilay Patel:** Is that reflected in your structure? You describe engineering and product and, uh, you know, your localization, but that piece, right? You've got to build that agentic orchestration layer. You've got to decide what models are best for which purpose. You could design an engineering organization around that problem specifically. Is that how you've done it, or is is that done differently? **[0:46:23] Sean Fitzpatrick:** Yeah, yeah, uh, we we have an engineering team that, uh, focuses on that, uh, around the planner agent and the assignment of the tasks to different, um, agents. **[0:46:33] Nilay Patel:** Is that where the bulk of your investment is, or is it paying the token fees? **[0:46:37] Sean Fitzpatrick:** Uh, you know, I haven't actually broken it out that way, so I couldn't tell you. You know, the token fees are certainly, um, an important part of the investment. The engineering is a huge portion of the investment. Uh, the, um, uh, the attorneys that we hire to review the output and tell us, you know, is this good or is this not good? That's that's a massive, massive import, uh, um, piece of the investment. So it's it's spread out over many, many different things, but certainly we spend a lot, uh, a lot of money on that particular issue. **[0:47:07] Nilay Patel:** What? What? Tell me about those attorneys. Do you hire attorneys to basically do doc review of the AI? Are they very senior attorneys? Are they are they moonlighting from big firms? There's a bunch of junior associates in a basement. **[0:47:17] Sean Fitzpatrick:** Yeah, it it's based on the task, right? So what we try to do is get attorneys that have experience in a par particular matter. So if we're looking at, um, documents related to an M and A transaction, you know, we want those to be looked at by someone who has some experience in mergers and acquisitions. And they can tell us that, yeah, that that document looks great, or, you know what it's missing, it's missing these particular things. **[0:47:40] Sean Fitzpatrick:** And then we can go back and say, why did we miss those particular things? And what changes do we need to make to the way that we're, um, training these models and directing these models, uh, to correct that situation going forward? **[0:47:53] Nilay Patel:** What's the biggest thing you've learned from that process? **[0:47:57] Sean Fitzpatrick:** Uh, I guess the the the biggest thing I've learned is how important it is to have attorneys, uh, doing that work. I mean, that was, you know, I expected to to hire a lot of technical people and data scientists to do this work. I didn't really expect to hire an army of attorneys. Um, but I think it's kind of one of the secret sauce, uh, components of our solution is that, um, our outputs are attorney reviewed, and so that's how we keep getting the, uh, more relevant, um, results. **[0:48:28] Nilay Patel:** Where were you best at to start with, and where were you worst at to start at practice area wise? **[0:48:33] Sean Fitzpatrick:** Uh, you know, I, I, I guess we we weren't really good at anything, uh, to begin with, right? And I and I I think we're, you know, we're kind of building things out. Sometimes it's a practice area, sometimes it's a a task. You know, I think that the there's a big focus, you know, if if you look at all those different tasks that we were talking about e- earlier that attorneys do, in many cases, the output of that task is some sort of a document, right? **[0:48:59] Sean Fitzpatrick:** And so we're really focused right now on, like, how do we improve our document drafting? **[0:49:04] Nilay Patel:** Um, is all this revenue positive yet? Are you are you making money on all this investment, or do you see that in the on the horizon? **[0:49:11] Sean Fitzpatrick:** You know, our growth rate has definitely accelerated, uh, as a result of this, you know, the the main thing that we're focused on is the the customer outcome. And so what we're seeing is the that the customers are getting happier and happier and happier with the solution. And so I would say that it's been, uh, very successful in that regard. And it's it's the fastest growing product that we've ever had. **[0:49:31] Nilay Patel:** But is it growing fast but losing money with every query? Is that right? **[0:49:34] Sean Fitzpatrick:** Yeah, yeah, no, we're we're not there. We're not losing money with every query. **[0:49:38] Nilay Patel:** Okay, are you breaking even? or are you making money? **[0:49:40] Sean Fitzpatrick:** It's, yeah, yeah, I mean, our profit is growing. **[0:49:43] Nilay Patel:** On specifically on AI tools or overall? **[0:49:46] Sean Fitzpatrick:** Yeah, yeah, I mean, most most of our investments in AI tools. **[0:49:49] Nilay Patel:** Great. Um, let me zoom out. Let me take the last bit here and just zoom out even more broadly. And I mentioned that I would bring up precedent again in this conversation. It I think if you're paying attention to the legal system in America right now, you know that it's in a state of pretty much pure upheaval, right? You've got district court judges calling out the Supreme Court, which is not a thing that usually happens. You have a Supreme Court that is overturning precedents in a way that I don't know, I I feel like I learned nothing in law school, right? **[0:50:18] Nilay Patel:** Chevron deference is out the door. Uh, Humphrey's executor, the law that keeps the president from firing FTC commissioners, I'm guessing, is out the door. Uh, Roe v. Wade, it was out the door. Like, just these foundational precedents are out the door. A lot of that is based on what conservative judges would call originalism. I have a lot of feelings about originalism, but a big trend inside of originalism is using AI or what they call corpus linguistics to determine what people meant in the past. **[0:50:49] Nilay Patel:** And then you take the AI and you say, well, it did the job for me. This is the answer. Are you worried that your tools will be used for that kind of effort? Because it really puts a lot of pressure on the AI tool to understand a lot of things. **[0:51:04] Sean Fitzpatrick:** I'm not that worried. I don't think the Supreme Court is is asking LexisNexis what we think they should do, and, uh, and then dis issuing opinions. **[0:51:14] Nilay Patel:** But certainly, but certainly courts up and down the chain are. **[0:51:17] Sean Fitzpatrick:** They're, yeah, they're they're asking legal questions, they're getting answers back, and then they're interpreting those answers. I think we're providing them with the raw content that they need to make the determinations, but we're we're not practicing law, we're not making those decisions for them. **[0:51:35] Nilay Patel:** We have to take on a short break. We'll be back in just a minute. **[0:51:37] Ad read:** This holiday, switch to the Simplicity plan from Verizon. It's one plan, one price, just $30 a line per month for you or the whole family. And on their best five G network, switching to Verizon and shopping their best deals makes the holidays simple. Head to verizon dot com or your nearest Verizon store today. Support for the show comes from Framer. The future of website building is seamlessly blending all the tools we have available. And you could use a platform that has it all and stays useful even after you launch. **[0:52:18] Ad read:** That's what Framer is. Framer is the AI website builder that helps creators, teams and businesses ship production ready sites faster than ever while getting every detail right. You're able to prompt, inspect, edit, and publish in one place at a whole new pace. On Framer, agents bring speed and scale, while people bring taste, judgment and control. And that all takes place in the same workspace where the real site is designed, managed, reviewed, and published. Everyone's ideas land on the canvas, stay editable, and can be published when the team is ready. **[0:52:51] Ad read:** Thousands of businesses, from early stage startups to Fortune five hundreds, are choosing to build their websites in Framer, where changes take minutes instead of days. Learn how you can get more out of your site from a Framer specialist or get started building for free today at framer dot com slash decoder for 30% off a Framer Pro annual plan. That's framer dot com slash decoder. Rules and restrictions may apply. Support for the show comes from Outshift, Cisco's incubation engine for frontier technology. **[0:53:26] Ad read:** You can find multi agent AI in every enterprise roadmap right now. But here's the problem everyone's facing. Agents can pass messages, but can't think together. These systems silently underperform through cognitive failures, misreadings, unverified claims, false consensus, and conventional monitoring can't detect them. Outshift by Cisco is building the fix with their new comprehensive resource, the Internet of Cognition. It's an open source guide that's a foundation for building multi agent systems from design to production. The goal is to give agents shared context, shared memory, and guardrails, all to drive the results we actually expect from AI. **[0:54:06] Ad read:** You can lean on this resource to understand agentic AI's evolution and how it shifted from isolated agents to orchestrated superintelligence. You can gain insights for how to future-proof your business by tackling complex challenges with multi-agent systems. You can read the paper, experience the demo, grab the code from Outshift dot com. That's O U T S H I F dot com. **[0:54:37] Nilay Patel:** Welcome back. I'm talking with LexisNexis CEO, Sean Fitzpatrick, about the wrinkle that AI is adding to originalism. And I kind of ended up springing a product demo on Sean to do it. I'm gonna spring this on you, but here it is. Here's John Bush is a Trump-appointed judge. He's he cited the emergence of corpus linguistics in the legal field. And he said, uh, to do originalism, I must undertake the highly laborious and time-consuming process of sifting through all this. **[0:55:03] Nilay Patel:** But what if AI were employed to do all the review of the hits and compile statistics on word meaning and usage? If the AI could be trusted, that would make the job much easier. Right? That is him saying I can outsource originalist thinking to an AI. And this is a trend. We I see this, particularly with the originalist judges, that the job they think they're they're meant to do is determine what a word meant in the past. **[0:55:26] Nilay Patel:** And AI is great at being like, statistically, this is what that word meant in the past. And we're gonna outsource some legal reasoning to them. And this is, I think, very odd. Like, uh, right, my thoughts about originalism aside, my thoughts about Stare Decisis in America in 2025 aside, saying I will use an AI to reach into the past and and determine this meeting seems very odd. And I'm just wondering how you feel about your tool being used in that way. **[0:55:52] Sean Fitzpatrick:** I definitely understand your point there. Um, you know, I I think about, like, the analogy of a brick, right? You can use a brick to build a hospital, take care of sick children. Or you could take a brick and you could throw it through a window, right? Uh, one use is really great, and one use is pretty negative. Um, but in neither case, it's it's a brick. I you know, I I I think about our tool as being, like, not good or bad. **[0:56:22] Sean Fitzpatrick:** I think it could be used for good, uh, you know, I think it could be used for, you know, any type of activity that that attorneys I I wouldn't wanna say originalism is a bad thing, right? I I think it could be used for many different things. I think it could be used for originalism. I think it could be helpful for those that that wanted to take that path and find a new way of looking at something. **[0:56:44] Sean Fitzpatrick:** Um, we have all the data. They can search it. They can use the tool to, um, find things that they, you know, it wasn't possible to find in the past. Uh, so I could see them, you know, using our tool in that way. Um, and I guess it's, you know, it's up to the attorneys to determine how they're going to to use the product. Um, we're not **[0:57:05] Nilay Patel:** Well, there's that. It's I mean, that's **[0:57:09] Sean Fitzpatrick:** we're not building it because we're trying to change the law, right? Uh, we're building it because we're trying to help attorneys do the tasks that they want to do. **[0:57:13] Nilay Patel:** Right, but I I look at the sweep of the tech industry, not the legal industry, but the tech industry over the past fifteen, twenty years. Boy, have I heard that answer many, many, many times. The social media companies all said, Well, you know, you can use it for good or evil. We're neutral platforms. And it turns out maybe they should have thought of some of these harms earlier. The AI companies today. Who knows if training is copyright? **[0:57:34] Nilay Patel:** Like, we know the answer. And you can't actually just opt out of copyright law. Like, and now we're gonna do the lawsuits and we'll see how it happens, right? Who knows if doing TikTok for deep fakes, like OpenAI is doing to Sora, is like, we actually, we know, right? Like, we we know the answers, and you should have some guardrails. So I'm I'm posing you the same question, right? We see a judiciary, particularly originalist judiciary, hell-bent on using originalism to change precedent at alarming rates. **[0:58:03] Nilay Patel:** I I would say for me, alarming rates only because I paid for a law degree, and now I think it's useless. That's why it's alarming to me. It's alarming because a lot of people have their rights taken away as well. Like, every day this is happening. And one of the tools they're going to use is deference to an AI decision engine. They're gonna say, We asked the AI, what did all people mean when the Fourteenth Amendment was drafted? **[0:58:25] Nilay Patel:** And this will be how we get to a birthrights decision case. And that I I'm just connecting it to the conversation we had at the beginning. We're gonna we're gonna we're gonna give our reasoning to a computer in a way that the computer is not necessarily accountable for, and we're gonna trust the computer, and that method of thinking and that rigor might go away. And so I, you know, I I've heard this answer, the tool is neutral, and so how I'll use it from tech companies for years, and I see the outcomes. **[0:58:48] Nilay Patel:** I'm asking you, you're building a tech product for lawyers. They're already using it in this specific way. And I'm wondering if you've thought about the guardrails. **[0:58:55] Sean Fitzpatrick:** Hm. We have responsible AI principles, uh, that we operate under, uh, and so that that includes a number of things. Uh, one is, uh, we always try to consider the real world implications of any product that we develop. Uh, we wanna make sure that there's, um, transparency, uh, in terms of how our product works. So we open up the black box so people can see the logic that that we're using, uh, and they can actually go and change it too if they if they want to, right? **[0:59:22] Sean Fitzpatrick:** So we wanna make sure that there's transparency and there's control. Uh, human oversight, right? So we always incorporate human oversight, uh, into the, um, development of our products. Uh, privacy and security is another one of our, um, core tenets of, uh, of responsible AI creation, and then the pro- um, prevention of the introduction of bias, um, is another thing that's that's incorporated in. So those are the relics principles, uh, uh, for AI development, and, uh, you know, we we adhere to those. **[0:59:51] Sean Fitzpatrick:** So, you know, we want to create products that do good things for the world. **[0:59:56] Nilay Patel:** If you ask Lexis AI if the Fourteenth Amendment guarantees birthright citizenship to all people born in the United States, will it make the argument that it doesn't? **[1:00:05] Sean Fitzpatrick:** I've never asked that question. I I can't tell you. **[1:00:08] Nilay Patel:** Do you have do you have a phone on you? There's a mobile app. **[1:00:09] Sean Fitzpatrick:** Oh, well, I could I could pop up here and ask it, I suppose. So, what was the, uh, hold on a second. Let me pop into Prodigy here. Alright, does the Fourteenth Amendment guarantee birthright citizenship, or are there exceptions? Hm, let's see. **[1:00:35] Nilay Patel:** Yeah, I'm very curious to see what it says because up until recently, there's only been one answer to that question. And now the Trump administration is saying there's nope, actually, that's not what subject to the jurisdiction thereof means. And they will, in order to win at the Supreme Court, have to construct an originalist argument to that question. And I am confident that the way they're gonna do that is they're gonna feed a bunch of data into an AI model and say, This is what the f- this is what was actually meant at the time of the drafting of the Fourteenth Amendment. **[1:01:05] Nilay Patel:** That is a th- that's a thing that AI will be used for that is very destructive. **[1:01:09] Sean Fitzpatrick:** I'm not an attorney, by the way, so I'm just gonna read the answer here. Fourteenth Amendment of the United States Constitution guarantees birthright citizenship to all persons born or naturalized in the United States and subject to its jurisdiction. The phrase subject to its jurisdiction has been interpreted to include nearly all individuals born on US soil, with a few narrow exceptions. These exceptions include foreign diplomats, uh, children of foreign diplomats, children of enemy forces in hostile occupation, children born in foreign public ships, and historically, children of members of Native American tribes who, uh, owed allegiance to their tribe rather than the United States. **[1:01:48] Nilay Patel:** Yeah, you should send that to John Roberts right now. You know, can yeah, so can Protagate do that? Because that's the answer. But the, you know, the question is, well, you said that on this podcast, are a bunch of conservative influencers gonna say Protagate is woke now? yeah, I mean, Right? Like, it it does this is the culture **[1:02:05] Sean Fitzpatrick:** recognize, yeah. **[1:02:05] Nilay Patel:** war that you're in. **[1:02:06] Sean Fitzpatrick:** It does recognize that, uh, recent cases have affirmed this interpretation, rejecting attempts to expand the exceptions of birthright citizenship. So it does also, you know, recognize that there have been efforts to, um, to interpret it in a different way. And and it goes the the, quite the answer goes on, you know, uh, quite quite a bit, so **[1:02:23] Nilay Patel:** Well, the reason I ask that question very specifically, that's that's the next precedent that's up for grabs. It's a it's a big one, it's foundational, right? That's reconstruction is up for grabs in a very real way in that case. Do you think as the toolmaker, you have a responsibility? Because that's really the the question for so many AI companies. You're the toolmaker. Do you have a responsibility to not deep fake real people? Do you have a responsibility to not show people fake ideas, right? **[1:02:49] Nilay Patel:** I think you're very clear on that. You have a responsibility to not hallucinate. But here you have a well, a quite. I'm wondering, and and the question and we don't want is **[1:02:55] Sean Fitzpatrick:** to, we don't want to, um, introduce or perpetuate any bias that might exist either. And so to do that, like, we, you know, rely back on the law, right? Uh, as opposed to, you know, like a a consumer-grade model that would just, you know, probably uses news articles, right? Which might have a very different interpretation of things, depending on which news articles, uh, and they're, you know, they're much more likely to be biased introduced into news articles than into black letter law, for example. **[1:03:25] Nilay Patel:** yeah, um, the reason I'm curious about that is, uh, you know, there's a spectrum. I think, uh, telling people what they can do with Microsoft Word running locally on their laptop, I I don't think there's any place for that. Uh, fine, do do what you gotta do. Telling people what they can do with a consumer-grade AI tool built into Facebook, I think Facebook has a lot of responsibility there, right? Especially cause the opportunity for them to distribute that content far and wide is is at their fingertips. **[1:03:51] Nilay Patel:** That's a big spectrum of opportunity. And I here in the middle with the AI companies, it's do you have the obligation to say, well, if you wanna go make the argument that birthright citizenship doesn't protect everyone in the United States, you gotta do that on your own. Like, our robot's not gonna help you. Do you feel any of that pressure? **[1:04:10] Sean Fitzpatrick:** Yeah, I mean, we try not to get into politics or any of the any I don't of that **[1:04:14] Nilay Patel:** think that's politics. debate, you know, I wanna be the a hundred I. D. I do not think that's politics. **[1:04:17] Sean Fitzpatrick:** yeah, uh, you know, we, you know, we're trying to develop a system that does not have bias introduced into it that will give you the facts, and attorneys can do the work that attorneys do and and make those important decisions. Our job is to give them the information that they need, the precedents, the the facts, uh, all the information that they need to then develop their their argument, whatever that might be. But we, yeah, we really don't get into, uh, any of the politics of, you know, birthright citizenship. **[1:04:48] Sean Fitzpatrick:** Is it is it guaranteed, or is it not guaranteed? **[1:04:51] Nilay Patel:** Well, it's at some point you do, right? I mean, this is, again, to to to bring us back where we started. I I first encountered LexisNexis as a database of cases. In in some case notes, right? There are some there are some law professors who are very proud that their case sense were in LexisNexis when I was in law school. Now we're drafting a little bit. Now we're gonna go do the research. Now we have agentic AI that's making the arguments. **[1:05:13] Nilay Patel:** Maybe one day we will automate all the way to filing. You're taking on more of the burden. You are making the arguments. The company is making the arguments. Where is the line? Because a a lawyer, there are lots of lawyers who wouldn't take that case, who wouldn't make that argument. Is there a line for you? **[1:05:30] Sean Fitzpatrick:** Yeah, I would say our a approach is to arm the attorneys, uh, with the best possible information, uh, help them with the drafting of those documents, and, uh, you know, we're really just being led by our customers and what they're what they're asking us to do. Uh, we certainly are not trying to interpret the law, um, we're not trying to shape the legal system. We're and we're not lawyers, right? We're not trying to do the work of lawyers. **[1:05:54] Sean Fitzpatrick:** We're trying to help lawyers do the work that they do in a more efficient way and and hopefully help them drive better outcomes. But it's always their prerogative to, you know, interpret the the information that we provide, which is what lawyers do, right? That's the what they're great at. And then, you know, the reason we have cases is because there are people on both sides of them, right? And that's they're making the two case two individuals are gonna make, uh, op- o- opposite arguments. **[1:06:19] Sean Fitzpatrick:** We wanna represent or we want to support both of those, uh, attorneys as as best that we can. So **[1:06:25] Nilay Patel:** Yeah, I I I get it when you're the database of cases, I get it when you're the word processor, I get it when you're the specialized word processor or the case management platform. The thing that I'm pushing on, sort of repeatedly here, is when the AI system is actually doing the work, do you feel like you have different guardrails? **[1:06:43] Sean Fitzpatrick:** the the work that we're doing, you know, I think our responsibility is to develop AI in a in a responsible way. Uh, and so I think, you know, we, that's what So, uh, we're trying to **[1:06:52] Nilay Patel:** just give me an example of something you wouldn't let your AI do. An argument that you wouldn't let your AI make, or a motion that you wouldn't let your AI draft. **[1:07:00] Sean Fitzpatrick:** I I don't know that we would wanna necessarily restrict the, um, the AI in that way. Um, you know, I think that the information in the system is, you know, we're we're referring back to the information that we have, like our authoritative collection of documents and materials, um, which helps lawyers understand what the what the facts are, what the precedent is, what the background is, um, and then they can do the real deep legal work and and make those trade-off decisions, the judgment decisions, those important, you know, things that, again, that you, you know, that attorneys went to law school to do. **[1:07:39] Nilay Patel:** I think these questions are they're gonna come up over and over again. We should have you back to answer them. But as you learn more, but as you look out over the horizon, the next two, three years, what's the next set of capabilities you see for LexisNexis, and what do you think the pressures that might change how you make some of those decisions will be? **[1:07:54] Sean Fitzpatrick:** Yeah, I think the main thing that's gonna change the, um, the path going forward, and it it's hard to say what exactly what it's gonna look like, because if I look back two years ago, I would have never guessed we'd be doing what we're doing today because the technology didn't exist or it was too expensive to to implement. Uh, that's totally changed over over the last two years, and I think over the next two years, it's going to change again. **[1:08:18] Sean Fitzpatrick:** Uh, and so it's really hard to say where where we're gonna go. You know, our vision remains the same, which is that we, you know, we wanna help attorneys, um, you know, we wanna provide them with a personalized, eh, AI, uh, powered product that understands their practice area, it understands their juris- jurisdiction, it has access to our authoritative set of materials, it has access to their, um, prior work product, it understands their preferences, it understands their style, understands what they're trying to do, and it can automate tasks that they do today manually. **[1:08:50] Sean Fitzpatrick:** And we'll continue to take that latest available technology, show it to our customers and have them help us understand how we can use that technology to serve them in more modern and relevant ways. And that's that's really the what's gonna guide our our roadmap in the future. **[1:09:06] Nilay Patel:** Well, Sean, this was great. Let me know when you develop a system that can actually navigate an electronic case filing website, cause some of the smartest people I know can't do that. But this was great. We gotta have you back soon. Thank you so much. **[1:09:16] Sean Fitzpatrick:** Thank you so much. Really enjoyed our time today. Take care. **[1:09:23] Nilay Patel:** I'd like to thank Sean Fitzpatrick for taking the time to join me on Decoder and thank you for listening. I hope you enjoyed it. If you'd like to let us know what you thought about this episode or really anything else, drop us a line. You can email us at decoder at the verge dot com. We really do read all the emails, or you can hit me up directly on Threads or Bluesky. We also now have a TikTok, an Instagram, and a YouTube. **[1:09:39] Nilay Patel:** They're all at DecoderPod and they're a lot of fun. If you like Decoder, please share with your friends and subscribe wherever you're podcasts. Decoder is produced by Chris Chibnall and part of the Vox Media podcast network. The show is produced by Kate Cox, Nick Stat, and edited by Ursula Wright. Our editorial director is Kevin McShea. The Decoder music is by Breakmaster Cylinder. We'll see you next time. **[1:10:40] Ad read:** As soon as autumn rolls around and you swap your iced lattes for pumpkin spice, that's your signal to start thinking about the holidays, the decor, the gifts, the shopping lists, but more importantly, the holiday treats, cheesecakes, cookies. Is there anything better? And when the recipe calls for cream cheese, there's only one option Philadelphia, the original cream cheese, made with fresh milk and real cream. This holiday season, whether you're going all out or keeping it simple, make it really, Philly good. **[1:10:52] Ad read:** Find your favorite holiday recipe at creamcheese dot com. Support for the show comes from Engine. Running a small business means every dollar has to work hard. But if your team is still booking travel the old way, it's costing you more than you think. Engine is the fastest growing travel and spend platform in the country, built specifically for businesses like yours. Book a trip in as little as two and a half minutes, earn up to 10% back on hotels, and in 2025, Engine customers saved more than $300 million on travel with zero booking fees, no contracts, and no BS. **[1:11:16] Ad read:** More than a thousand businesses join Engine every month. Join them and get $500 when your business signs up and starts traveling at engine dot com slash vox. **[1:11:28] Speaker 5:** You're either all in on this model, or maybe you're building with another. You're either speed or is it security? Or your AWS AI. Don't pick a side. Pick em all. AWS AI is how.