This week we catch up with Jeff Pfeifer and Serena Wellen from LexisNexis to discuss the rapid development of AI tools for the legal industry over the past year. Pfeifer and Wellen give us an insider’s view of what it took to bring their Lexis+ AI tool to the market and the balance between speed to market and getting solid customer guidance on what they need in a legal-focused Generative AI tool. Between the initial version released to a select group of customers and the current version, the product grew from an open-ended chat interface into more of a guided resource that helps users on creating and following up on prompts. As with most AI tools created in the past year, there is still more potential as more and more customers use it and give critical feedback along the way.
In addition to Lexis+ AI, LexisNexis has now launched two additional AI products – Lexis Snapshot and Lexis Create. Lexis Snapshot summarizes legal complaints to help firms monitor litigation. Lexis Create brings AI capabilities directly into Microsoft Word to assist with drafting and research while lawyers are working on documents. The goal is to embed insights where lawyers are actually doing their work rather than separate AI tools.
While the focus of the initial Generative AI tools from LexisNexis were focused on the US market, Serena Wellen and her team are busy expanding that to more of an international reach. This requires adapting the models, content, and interface to different languages and legal systems. This is complex undertaking, but Wellen discusses how LexisNexis has content and editors around the world to help customize the tools. Surprisingly, desired use cases are fairly consistent globally – both simple legal tasks as well as more advanced legal research and drafting.
Greg Lambert brings up a recent LinkedIn discussion that he had with Microsoft’s Jason Barnwell, where Barnwell told him that today’s version of Generative AI tools are “the worst these things will ever be.” In response, Pfeifer says that LexisNexis is focused on continuously improving answer quality to build trust and prove the value of AI to skeptical lawyers. LexisNexis is leveraging relationships with companies like Microsoft to reinforce the stability and progress being made.
Wellen and Pfeifer look into the future and predicted that AI assistants will become highly personalized to individual lawyers. AI agents will also proliferate across platforms to help automate tasks and workflows. Law firms will likely accelerate their adoption of AI tools based on rising expectations and demands from corporate legal departments to work more efficiently.
Twitter: @gebauerm, or @glambert
Threads: @glambertpod or @gebauerm66
Music: Jerry David DeCicca
Marlene Gebauer 0:07
Welcome to The Geek in Review. The podcast focused on innovative and creative ideas in the legal profession. I’m Marlene Gebauer. And
Greg Lambert 0:14
I’m Greg Lambert. So with the holidays coming up, we’re probably going to go on a little short hiatus toward mid December, I’m going to be yes, I’m going to be traveling with my lovely wife to Europe to celebrate our 30th anniversary. And I promised her I would not be editing podcast while we were on our river cruise up the Rhine. So
Marlene Gebauer 0:37
yeah, you should not do that.
Greg Lambert 0:39
It’s the little things that keep you going for 30 years just like that.
Marlene Gebauer 0:43
Yeah, and I will be I will be busy holiday shopping and preparing for the holidays. So we I think it’s important, we kind of keep that sanity. Just sort of keep going for next year. Yep. But you know, we do have some great guests lined up for early December and for 2024 already. So thanks to everyone that listens to The Geek in Review. We really do appreciate it. When we run into listeners while we’re traveling to conferences and events. I know, I will probably be seeing a few of you. And in the next week or so. Because Greg and I are going to be in New York, I guess. You want me to say next week or this.
Greg Lambert 1:21
It’ll be this week. By the time this comes out. Speaking of speaking of which, Greg
Marlene Gebauer 1:25
and I are going to be in New York this week at the legal AI Pathfinders Assembly in New York, where we’ll be talking to Dan Katz about how far we’ve come with generative AI in the past year, I will also be speaking with my co workers on a panel about the AI hype cycle. And I’m going to be in Miami next week at the TLFT conference and hopefully get to speak with somebody there.
Greg Lambert 1:48
I’m really looking forward to the discussion that we’re going to have in New York. And speaking about just how far we’ve gone with generative AI in 2023. I think as we’re recording this, we’re like right at the one year anniversary of the public launch of ChatGPT, which I think will we’ll be looking at the world it’s pre ChatGPT and post ChatGPT. Now, so but total difference a year makes. It does it does. And I think today’s guest epitomize that the topic and you know, everything that LexisNexis has accomplished this year, and getting their generative AI ai tools up and running.
Marlene Gebauer 2:31
So we’d like to welcome Jeff Pfeifer Chief Product Officer, Canada, Ireland, UK and US at LexisNexis. And Serena Wellen, Senior Director of Product Management at LexisNexis. Jeff and Serena, it’s very good to have you here on The Geek in Review.
Serena Wellen 2:47
Great to be here.
Jeff Pfeifer 2:48
Great to be with you.
Greg Lambert 2:49
Yeah. Thanks for joining us, Jeff and Serena. So we brought you on the show to talk a little bit more about the launch, not just of the Lexis plus AI tool. But it’s really expanded on that by adding a couple of additions to the AI products, which are integrated into the product that every lawyer lives in. And that’s Microsoft Word, which are the new Lexis Snapshot and Lexis Create.
Marlene Gebauer 3:16
And, before we get into the details, I wanted to stop for a second and just think, you know, a year ago, barely anyone was talking about Gen AI tools. Yet for the past 11 months or so, this has got to have taken up most of your waking moments. You know, Jeff, and Serena, how is 2023 been for both of you?
Jeff Pfeifer 3:34
Well, it’s a it’s a great question, because this year has really been an interesting one. When I think back a year ago, obviously, for those of us that work in the technology space, were aware of what was happening more broadly and had even been experimenting ourselves. But I don’t think anyone could have predicted how quickly the explosion and use of of ChatGPT in particular could have happened. And yet at the same time, I think back over the last 12 months, how quickly the entire legal profession is really embrace the technology and experimentation with the technology. I think about just how many firms that we work with have active experimentation underway. I often laugh that, you know, law is thought of as not an area to quickly embrace new technology. But in this space, the last 12 months has really been remarkable by how quickly people have embraced the capabilities. And as I’ve alluded to you before, I think that’s because it’s so closely matched to what lawyers do every day. And that creates a really nice marriage between the technology and what the possibilities are for the legal market.
Marlene Gebauer 4:46
And Serena, how about you?
Serena Wellen 4:48
Yeah, I it’s, it’s been quite amazing. I and my team have worked with technology for a really long time. And some of the problems The technology problems that we wanted to solve for years. One example is like auto summarization are suddenly within reach immediately. And there’s so much to learn. New developments are coming almost daily there for a while, flip down slightly. But it’s just been a really fast paced, exciting time.
Greg Lambert 5:23
I think we’ve we’ve had a little bit of a speed up the last few weeks, at least some of the news going going around as well. So, but Jeff, we had you back on in May cashes feels longer than that, doesn’t it? To talk about the customer preview program that LexisNexis was rolling out to solicit the feedback for the Lexis plus AI. And full disclosure, my firm at Jackson Walker we participated in that we kind of we kind of blocked in to Merlin’s team did as well. And so, you know, I, I’d be interested in learning, you know, how everyone else’s feedback influenced, you know, the final decision, because I think what we saw in April and May and June, is just a shell of what we’re seeing now. So you know, what major changes from the initial announcements did the final product turnout, it
Jeff Pfeifer 6:24
was an interesting journey. If we rewind the clock, you know, we had really clear ideas about what we thought people could do with the technology when we initially launched the product. But the purpose of the whole engagement preview program was really to learn how people actually use the product once it was live. In some ways, it was quite different than what we had imagined we originally conceived somewhat similarly to the way ChatGPT works, that we would have just a very open ended bots interaction that would allow people to virtually ask anything, but we found that our customers struggled with that a little bit, they actually wanted to have more direction about what capabilities were supported by the technology. And ironically, that desire on their end helps us out a lot, because if they can signal to us what they’re trying to do, we can make sure that we apply the appropriate Large Language Models, technology calls to retrieve the best possible results. So from the preview, we really learned a couple of key things. First, we introduced some task organization to the product so that people could signal to us what they were trying to do. And we use that information to improve the quality of the answer results, we learned that they wanted to have the answers that our model generates grounded in a wider set of legal documents. So we’ve been working hard to incorporate more and more grounding data as we refer to it LexisNexis trusted content that forms the basis of the answer. And we also learned that they had a strong desire to participate in what we now refer to as multi turn interactions, the ability to clarify or to reframe a question after they’ve looked at the answer that’s been generated by the model. And so we’ve incorporated that capability now, along with some other ones that we’ve learned about, like uploading and summarizing information. But overall, it was a tremendously collaborative process where we got great insights about how users actually use the product.
Greg Lambert 8:40
That was one of the things that I’ve talked with many vendors about is that the thing that I think kind of catapulted ChatGPT, and made it such a big hit so quickly, was the interface was that it was a chat. It was you know, you could take a question, and then follow up on it. And I think the initial round of products that we were seeing, were kind of one shot, you know, you put in one and then you kind of have to start over again. And so by I think including that magic into the legal products as well. I know it wasn’t easy, but I think it makes it a lot more interesting to work within and gives you that chit chat capability. Is that kind of what you were getting in your feedback. Yeah,
Jeff Pfeifer 9:26
absolutely. And we’ve talked previously about one of the long held problems in search is that it is always one shot. You ask a search question you get one shot answer or documents listed. Clients have always wanted to ask follow up questions to ask clarifying questions. And to me, that’s one of the most exciting things about the technology is now we have this opportunity to create what amounts to a human like dialogue between the user and the service itself.
Marlene Gebauer 9:58
Serena, you know, what are the key key factors in maximizing the effectiveness of Gen AI ai is prompt generation. And I can tell you this, this has been a big discussion at my firm in terms of what people need to know how to approach people about learning about prompt generation, because it is different than search. You know, people are not quite used to it yet, and how to use it. So how are you helping customers and law students and law librarians and others learn how to navigate the drafting the prompts needed for Lexis plus AI,
Serena Wellen 10:34
I agree that this is a really critical piece in terms of shifting behavior, and helping people get the most out of generative AI systems, including ours. So the place that we start with users of Lexis plus AI is that it’s a conversational system. So kind of echoing what Jack just said about multi turn. Large Language Models are really good at understanding and generating language. So like you and I, they benefit from getting more information about what you want about what the user wants. And there are a couple of mechanisms to do that. In Lexis plus AI. One is prompting, and the other is multi turn interaction. Of course, they’re intertwined. And since this is new to everyone, we are definitely training our users on what constitutes a good prompt. So how to start their conversations with Lexis plus AI, with what we call a well formed question, request, or instruction. And I think the thing to keep in mind here is, it’s good to say more, to identify necessary context to provide jurisdiction to include concepts and facts. The things that might be considered noise in a traditional research or search based system up to the present are really important when you’re putting together a prompt, add. Another thing that we are doing in Lexis plus AI, is we’re applying user interface intervention. So things like buttons or suggestions for next steps in the conversation. And what this does is support people within the product, help them easily get more familiar with the kind of iterative interaction that you can do with like, plus
Marlene Gebauer 12:33
AI. So I mean, we’re talking about this is Lexis plus AI. But Jeff, you’ve also recently launched two other products that use generative AI ai Lexis snapshot and Lexis create. Can you tell us a little bit more about those?
Jeff Pfeifer 12:47
Sure. So you know, I think for the last six months or so we’ve been talking about AI specific products, I hope a year from now, we’re seeing AI embedded in lots of different places, it just creates wide value, and we talk less about the fact that it’s an AI specific product, and more about how it enables certain work capabilities that actually underlies both of those two product launches that you referenced Marlene, the first is Lexis snapshot, and it builds on something that we learned about in our discovery work that summarization was really a great use case for the legal market. And so with Lexis snapshot, we have started by summarizing all federal litigation complaints that are filed civil complaints. And we provide that information to anyone who has an alert on either a litigant or a specific nature of suit with our court lead service, and that enables firms in a variety of different use cases to quickly scan the dates dockets learn if an entry or an update is important to them, or something that they should follow up on. And we see that potentially as applied to many different content types, not just complaints that are filed in court, but perhaps proposed legislation, proposed regulations, other types of information where there’s a need to quickly get a landscape view of what’s happened. And then action follow up activity from that. So the snapshot service will expand over time. To provide more of the summarization capabilities. We’ve learned a lot since we made it available to all of our customers under a free trial through the end of the year, that they’d like to see some additional information in the summary, some additional insights that they might be able to extract out of that summary. So I’m really excited about that space. Again, I think it’s a great marriage between the technology and something that our customers are trying to do every day. The second Lexis create is something that we launched a bit ago. It is a set of tools that help people in drafting actively in Word, and we recently announced wants the ability to bring these Large Language Models capabilities, the ability to ask a question, or draft a clause, for example, natively into Word. And so while one is actively updating or revising a document, they can use the generative AI ai capabilities from Lexis plus AI to bring those insights directly into the workflow experience in Word. And I’ve long been a believer that ultimately, that’s what our clients want. They want insights from LexisNexis. But they want them where they’re working. And as gray alluded earlier in the conversation, that is generally work, and people spend most of their time they’re actively creating work product. So we’ll continue to build out that capability and provide more insights at the point of workflow in a place like Word or Outlook or somewhere else that a customer may be working. Yeah,
Greg Lambert 15:57
I agree with you on the I think eventually, we will stop attaching AI to whatever it is, and it will just be embedded. The way that I’ve looked at it this year is if you go back, I think to like 2002 to 2006, somewhere around there, when the natural language searching first came out. To me, this is finally the true natural language searching that everyone hoped for, in the early days.
Jeff Pfeifer 16:25
Yeah, and then not said at this stage of the game, we really think it’s important to disclose that it is AI generated. We have we have long talked about responsible application and safe application of AI in various product applications. So in our products, we clearly label information that’s generated by AI as generated by AI so that users are aware of where it’s being used, at least for the foreseeable future, that transparent application of the technology is really important, because we know that we need to build trust with clients that they can rely on the information that the quality continues to improve. So that eventually I believe they’ll see that as a generated content, just like they would expect from any other LexisNexis brought. And
Greg Lambert 17:17
then we can generate a letter to the Fifth Circuit telling them what generative AI we used to enter in our process and how we verified that. Well, Serena, I know there are plans to expand Lexis plus AI internationally, which can’t be an easy undertaking, I would imagine. So how is it being tailored to accommodate the you know, the various language systems and legal systems that are not everyone is the same as the US, not just in language, but in culture and practice of their legal system? So how do you even approach going international like that?
Serena Wellen 17:59
It’s a really large and complex undertaking. But if I had to boil it down into three things, there are three things we’re doing to localize Lexis plus AI for use in other countries. The first thing is, we are changing the content that we use for grounding, or retrieval augmented generation is, is our ag is is also kind of a term of art, an industry term that’s used for the content that the model consult, when it is responding to a prompt and user prompt. So we’re changing that grounding, or our ag content to the local content. Second, we’re also adjusting the fine tuning of the models that we use, through prompt engineering to reflect what you just said, the local language and the nuances of the legal systems in other countries. And then where we need to we are tweaking the application itself to fit the needs of users and other countries. And I, you know, just a comment here in doing this. While it is a really complex undertaking. We benefit from having excellent content coverage, including practical guidance content. For other geographies, we really are a global enterprise. And we can also leverage the really robust teams of local legal and content experts that we have in place across the world. So the editorial and other teams that are working on all of that practical guidance and other content. Those are subject matter experts. And their expertise is really critical in ensuring that the model output fits the needs of local lawyers. So we don’t have a US lawyers, for instance, trying to guess that or interpret what are the needs of lawyers in the French legal system? them, or lawyers in the in the UK or, or Canadian legal system. And
Jeff Pfeifer 20:05
there have been some funny surprises along the way, which I’ll tell you a quick anecdote story. We had a team of users internally that began asking questions of Lexis plus AI in Spanish. And as the system is designed to do it retrieved documents that were written in English, the model consumed those documents for presentation of an answer, but translated the answer into Spanish for the user because it understood contextually that the original question was asked in Spanish. So while we haven’t, you know, release that as a capability broadly, I think it reflects, actually, again, a really core capability of models that they do translate well across languages. And so we’re really eager and interested in seeing how can we use benefits like that to globalize product solutions like this and make them available to users in many languages or to consume content from other jurisdictions written in another native language, but perhaps presented to a user in their native language?
Marlene Gebauer 21:11
That’s pretty cool. I was going to ask I mean, are you seeing different trends like across jurisdictions in terms of how people want to use the product? It’s funny,
Jeff Pfeifer 21:22
we did a wide survey on this globally. And we found that the use case interest was actually quite aligned, there were what I would refer to as two broad categories. So we’ve continued to survey individuals in legal markets around the world. And the first bucket of activities I sometimes refer to as mundane tasks, you know, simple drafting activities, like write an email, or draft a client response to something. The second category would be more substantive legal questions, things like legal research, or drafting a legal memo, things like that. And we saw actually fairly consistent and uniform interests, while it might unbalance skew a bit in some markets to things like contract drafting, because of the nature of the legal market, versus in the US, it might up on balance. It’s focused on areas like argument drafting for litigation. But we did see generally speaking, some fairly consistent interest in the same type use cases around the world.
Greg Lambert 22:25
Well, I’m excited because I know some some of our listeners are really hoping to do some research on in, cling on, and I’m sure that they’ll, they’ll be able to pull that off now.
Serena Wellen 22:39
We’re gonna try that.
Jeff Pfeifer 22:40
That is not a supported language.
Marlene Gebauer 22:42
Elvish… what else?
Greg Lambert 22:45
Depends on what their what their fetish is. So? Well, Jeff, when I was online a couple of weeks ago, and ended up having an interaction with Jason Barnwell from Microsoft. And he told me something that I kind of repeated a few times because we live, you know, a lot with issue spotters, that the lawyers are trained. So Jason had told me, he said, when he talked about AI products that exists right now that and his quote was, this is the worst these things will ever be. And, you know, in other words, you know, most of us believe that these tools will continue to show progress and improvement on a monthly even even daily basis. So, Jeff, what do you tell your customers, most of whom, like I said, are professional issue spotters? Who might say that the AI products just aren’t good enough for them to use in their own practice? And how are you leveraging the relationship that you have with with players like Microsoft and others, to share your customers, the, you know, the stability and continual improvement? It’s
Jeff Pfeifer 24:01
a really great question. I love Jason’s comment for so many reasons, you know, think of broad technologies. If somebody had said, you know, analog television in the 70s, you said, this is the worst it’s ever going to be, you’d say, it was consumable, we could use it and it added value. Today, is it much, much better? Yes, it’s infinitely better than what we could have imagined then. And it seems to me we have similar parallels here. But with one huge difference, the rate of improvement in this technology space is faster than I’ve seen in any other technology category, in my long tenure in developing product solutions for the legal market. So to Jason’s point, the quality is at a starting point, which our customers actually tell us right now is pretty good. They really like what it does and how it gives them an advantage in starting activities. But that rate of improvement is going to be significant. Can’t. And I would say that that users will see that in an iterative way. And we love issue spotters, because issue spotters tell us where we have a need to focus on a particular problem resolution. So if they say for example, as they have told us that they would like to see more citing authority, they would like to see more diverse citing authority in the answers that we generate, that allows us to experiment on ways that we bring additional citing references to the product experience itself, they’re telling us a lot about where they’d like to see more summarization. So they don’t have to click through and read every underlying document. We take that feedback and tried to iterate as quickly as we can to make improvements into the product experience. But my sense is that from the engagements I’ve had with clients, they’re pretty surprised at what the starting point looks like now, and can imagine what that iterative development will deliver over the coming weeks and months. So
Marlene Gebauer 26:03
Serena, this is clearly an exciting time to be innovating, right? What’s the atmosphere like in the product development team at Lexis? I’m imagining it’s very high energy right now. And you know, I’m also curious if you’re hiring for new and different skill sets than you were before? And how do you plan to keep up with the developments that seems to be happening like every day?
Serena Wellen 26:28
So you’re right, Marlene, is a really exciting time to be a product manager at Lexis. The work is so interesting, nothing is better than working on products that are truly transformational, and solve really critical problems for our customers. That’s really why we become product managers. It’s what drives us. So to be able to work with such powerful technology and really solve critical problems for our customers like summarization, like drafting and improve outcomes for them. That’s really exciting. As for my recruiting for product managers, I would say no, you know, I’m hiring for the same skills that I’ve been looking for in product managers for the past four or five years. So looking for AI familiarity, comfort with technology, or technology, acumen, customer focus, really critical creativity and the ability, or maybe I might call it drive to learn really, really fast. These are the attributes that I look for in product managers when I’m hiring and and I happen for a while. So we’ve been delivering AI based products for several years. So I would say this is not new for us. And then you asked about how do we keep up, it is a really fast pace, in terms of the technology changes and how we need to address them. So there’s lots of ways that we’re doing this in terms of internal education. But one of the ways that we are trying to foster broad innovation that I really like. And what I’m seeing in terms of how my team is responding, is we’re getting employees direct access to the models, through a console so that they can test and develop their own ideas. And it’s quite exciting what they’re doing. So they’re they’re doing quite a bit of product engineering, to solve problems that they’re working with in their own workstream.
Marlene Gebauer 28:36
That’s a fantastic idea. That’s really, that’s really cool. I guess I have a follow up question. In terms of, again, there’s so many different avenues you can take in terms of how to focus on the new development, like there’s so many different development, where do you decide like, Okay, this is we’re gonna, we’re gonna put our time and effort. Maybe we have to maybe not put as much time and effort into a sort of another area, you know, how do you make those decisions?
Serena Wellen 29:06
We have a strategy and a vision for AI at Lexis and what we want to be able to do for our customers. So that’s our framework for prioritizing work. And then we really look at how critical something is for the customer. What kind of impact what kind of value are we going to be able to deliver through the application of this technology, for instance, with something like summarization, this has tremendous impact for the customer. But it also has a lot of business impact for us. Because we have so much content and we want to be able to connect the customer to that content in a way that’s really going to accelerate their ability to consume it. And so something like summarization is a place where we want to spend a lot of time with generative AI to be able to do things like snapshot, to be able to do things like upload and summarize. And to be able to bring summarization across our product suite,
Greg Lambert 30:09
Jeff and Serena, Marlene and I have both contributed to it’s been a while it’s been pre pandemic, but to the advisory boards that Lexis pulls together to get input from people like km professionals and librarians and IT professionals. I know things have been, you know, just hyper focused on AI this year. But are you still integrating those results that you get are those inputs that you get from these advisors to help you kind of determine where you need to be headed?
Jeff Pfeifer 30:47
Absolutely, I mean, more broadly, our entire product strategy, we like to refer to it as lead through customer discovery process. So finding where our clients find the most important areas for focus, we obviously are still doing development in a lot of areas outside of what’s happening here. generative AI, for example, this year, we launched interaction in the cloud, new product called interaction plus. So there’s tremendous amounts of development taking place throughout the organization still informed by what our clients are telling us they’d like to see. But what’s interesting is that this year, in particular, our clients are telling us about their experimentation in the generative AI space. So almost every conversation leads back to what they’re exploring internally, and where we might have an opportunity to either co develop or support or explore things that are important on both sides. So back to our leads question, I mean, in any of those spaces, we’re looking to experiment quickly failed fast, find out what works and see if we can build it into a bigger solution. And that might be in ideas related to Lexis plus AI, but it might be related to ideas about bringing new and valuable summarized content into a place like interaction plots, because someone wants to learn more about a prospect or about a client that might be involved in in specific litigation activity. So across the board, I think the mantra still applies focus on those priority areas that our customers tell us are important to them, and important to business problems that they’re trying to solve. With there
Greg Lambert 32:30
being just so much buzz around generative AI, I’m glad that you brought up, you know, that’s not the only thing we’re working on. But it’s you know, it seems to be that all we’re talking about, and again, with the I think Serena mentioned this earlier, in that, you know, typically the legal industry is slow to adopt the new technology. But it’s been, at least on the surface, it seems to have been very active and determining what generative AI products are going to be useful based on the lawyers and the culture and the vision of the individual law firms. So Jeff, let me ask you, so how do you then approach these leaders and talk about how Lexis is differentiating itself from others and cutting through the hype? And when you go and speak to the leadership? How do you convince them that Lexus plus AI and your other tools are things that are really relevant to the solutions that they need?
Jeff Pfeifer 33:35
A couple of quick thoughts? You know, first, I had this great opportunity about a month ago, to participate in a Wall Street Journal conference called Tech live that happens every year in California. And it’s a bit of a who’s who, across all industries in technology. And one takeaway I had from that was that legal, as you eluded, Greg is well ahead of many, many vertical industries. And I talked to leaders from other industries where you would expect this development to be much further ahead. And they were amazed at how many commercially available solutions are already available in the legal market, including the one that we developed. And the takeaway from that is, again, I think there’s a really strong marriage between the technology and what lawyers do every day, which is driving the accelerated pace of product development that we’ve seen. So when we talk to clients, I’d say the biggest thing that’s driving fast action and activity here is demand on the corporate side. So every in house department is talking to the law firms they work with, about what their strategy is regenerative AI, how it might improve the quality of work product are the piece of work product that’s developed on behalf of the corporation. And as we look forward to the future, you know, I really see that as the number one driver of what’s forcing the pace of adoption and experimentation in the legal market. And I expect that to continue. So for us, I would hope again, fast forward 12 months from now, I hope what really distinguishes generative AI offerings is an assessment of the quality. So we’re talking less about specific models, and we’re talking less about the actual technology implementation. And clients are looking at our product through the lens of whether or not it helps me do things faster, the answer, quality is high. And I can connect the dots on things that I might not have been able to do otherwise. So that’s what we’re really focused on reviewing things like answered quality, the usefulness, the structure of the answers that are generated by our AI systems to ensure that they’re really adding value, and that firms have a way to measure the ROI from the investments that they’re making in tools like ours.
Marlene Gebauer 35:58
Well, we could talk offline about like who we are ahead of in terms of innovation, I’d be curious to hear about that. We’re at the point where we have our crystal ball question, and I will pose it to both of you. How do you see AI products impacting the work of law firms, the relationships between law firms and their clients? How are using these products going to change that?
Serena Wellen 36:25
This is an interesting question, because I think that, as Jeff was saying, law firms are really going to be driven by their clients to explore these avenues very rapidly, and to iterate on their own experimentation and trying to find ways to accelerate their own workflows in service to their clients. And one area I think that will be particularly important is that every organization is going to be able to securely I believe, use Large Language Models capabilities, with their own data. So to summarize, to analyze, to generate documents, and and do a variety of very sophisticated tasks with their own data without then having to train and deploy their own models. So I think this is coming. This is an area that we’re working on. And I think it’s going to benefit both law firms and their clients and their clients are going to be very interested in this. I also think that there’s going to be a proliferation of ai, ai agents. So as an example, we have an AI assistant in Lexis plus AI, Microsoft’s co pilot, another example of an AI agent, we’re going to have a lot of these that are going to be created that are going to be able to retrieve information and take actions across different products, different platforms and systems, the sophistication of this activity, and the ways in which these agents can interact with the systems and eventually each other is where I think we’re going to see a lot of time savings and productivity gains for lawyers. So more to come on that.
Marlene Gebauer 38:22
All right, Jess, what do you think
Jeff Pfeifer 38:25
I well build on what Serena said. And I think the future is really one where these agents, these AI assistants are highly personalized. And so that agent will have good insight into how you like to write, how you’d like to communicate, perhaps give you a proactive advice on issues that it knows based on your documents and what you’re working on are important to you. That is entirely possible. I’m a little distressed more broadly, right now, about the discussion of these tools and services. As chat bots, I don’t really think that does a good its service to the technology capability. We’re where we’re really headed as an industry. chatbots are obviously framed based on most people’s interaction with, say, a service organization. Agents is really the right way to think about it as as Serena positioned. It really is something that can learn from your interactions, advise and provide new information and insights to you as you’re working. And in that way, if you might imagine the future you’d have a service capability that behaves alongside you, much like a trusted professional colleague that you might work with today. And that might seem difficult to get your head around. But I think that’s very possible. And we’re probably looking at that capability within the next two to three years. That’s not far off into the future. So whether it’s the copilot capabilities that Serena referenced Then we’re working to develop, or Amazon’s new queue service, which was just launched this week. It’s really about personalizing that interaction experience between the user and an AI service in a way that adds more value much again, like you would get if you were working with a professional colleague.
Greg Lambert 40:18
Well, great, great insights. And I think what what Jeff just did, Marlene was invited himself back, probably in six months to talk about whether or not what he’s seeing is going to come true.
Marlene Gebauer 40:33
I don’t know. Where are we at now? Are we at three or four?
Jeff Pfeifer 40:37
I think four.
Marlene Gebauer 40:39
Oh, my God, you’re almost ready for the crown.
Greg Lambert 40:44
Or at least a smoking jacket.
Jeff Pfeifer 40:46
Cut moon has that right?
Marlene Gebauer 40:47
Yes, that’s right.
Greg Lambert 40:50
All right. Well, Jeff Jeff Pfeifer and Serena Wellen from LexisNexis, thank you both for coming on the show and talking with us. It’s great to be with you.
Marlene Gebauer 40:58
And a pleasure. And of course, thanks to all of you, our audience for taking the time to listen to The Geek in Review podcast. If you enjoyed the show, share it with a colleague. We’d love to hear from you. So reach out to us on social media. I could be found on LinkedIn, or on exit gave our M and on threads at M gave our 66 Yeah, and
Greg Lambert 41:17
I can be reached on LinkedIn or on x at glambertpod or on glambertpod threads. So Serena and Jeff, what’s your preferred place for people to reach out to you online,
Serena Wellen 41:30
Please contact me on LinkedIn.
Jeff Pfeifer 41:32
So you here can find me on LinkedIn for sure. That’s
Marlene Gebauer 41:35
good. Keeping it simple. I like it. And as always, the music you hear is from Jerry David DeCicca. Thank you so much, Jerry.
Greg Lambert 41:44
Thanks, Jerry. Alright, so talk to you later.
Marlene Gebauer 41:46
All right. Thanks, everyone. Bye.