Latest / The Adults In The AI Room / Earley AI Podcast - Episode 84: AI in Legal Operations with Mike Anderson

Transcript

Earley AI Podcast - Episode 84: AI in Legal Operations with Mike Anderson · The Adults In The AI Room

• 0:01
  SPEAKER_01: This is the Early AI Podcast hosted by Seth R detailed deep into the passions, expertise, and experiences of thought leaders and practitioners to talk about what's possible for artificial intelligence.
• 0:14
  SPEAKER_01: The Early AI Podcast is sponsored by Early Information Science, your digital transformation journey with design and deployment of innovative technology solutions as well as FACTAR.
• 0:26
  SPEAKER_01: Now enjoy the shop.
• 0:28
  SPEAKER_02: Well, welcome to today's podcast.
• 0:30
  SPEAKER_02: My name is Seth Early.
• 0:32
  SPEAKER_02: I'm your host, and on the Early AI Podcast, we talk about how organizations are getting value from AI, how they're running into problems and solving those problems, how artificial intelligence is really reshaping how organizations manage information, process information, produce value for customers, produce internal value, and be able to really transform the enterprise.
• 0:54
  SPEAKER_02: Today we're going to be looking at one of the real most demanding real-world environments for AI, the legal operations and litigation workflows.
• 1:02
  SPEAKER_02: This is a domain where accuracy is critical.
• 1:06
  SPEAKER_02: We can't, it's not an option.
• 1:07
  SPEAKER_02: Context is everything, and trust is also not an option.
• 1:10
  SPEAKER_02: We have to be able to get these things right.
• 1:12
  SPEAKER_02: And joining me today is Mike Anderson.
• 1:14
  SPEAKER_02: He is chief product officer at File Vine.
• 1:17
  SPEAKER_02: Mike has spent many years helping legal organizations modernize how they manage cases, documents, deadlines, and workflows.
• 1:24
  SPEAKER_02: In this conversation, we're going to explore what it takes to bring AI into the legal world in a way that's very usable, safe, valuable, and what other industries can learn from those constraints.
• 1:35
  SPEAKER_02: Mike, welcome to the program.
• 1:37
  SPEAKER_00: Hi, Sal.
• 1:37
  SPEAKER_00: Thanks for having me.
• 1:38
  SPEAKER_00: Glad to be here.
• 1:39
  SPEAKER_02: Terrific.
• 1:40
  SPEAKER_02: So what I'd like to start with are misconceptions.
• 1:43
  SPEAKER_02: What do people not understand about this world?
• 1:46
  SPEAKER_02: What things do in AI today in general?
• 1:49
  SPEAKER_02: What do you run into?
• 1:50
  SPEAKER_02: And then in the legal for legal teams and the legal operations.
• 1:54
  SPEAKER_02: And when you talk to firms, what are the biggest fears or misunderstandings that they have?
• 1:59
  SPEAKER_02: Whether that's the things that you have to address up front, whether it's replacing their jobs or hallucinations, data risk, and so on.
• 2:06
  SPEAKER_02: So let's talk a little bit about that environment.
• 2:09
  SPEAKER_00: Yeah, there's a lot to say on those topics.
• 2:11
  SPEAKER_00: Um, so I'll start with one of the misconceptions that I hear in kind of the AI tech industry chatter about the legal industry, and then we'll talk about kind of other misconceptions.
• 2:22
  SPEAKER_00: Yeah, one of the first things that are often that it's often talked about is okay, could AI take the job of an attorney or take the job of a paralegal?
• 2:32
  SPEAKER_00: And it's always interesting to me when people make this point because I want to ask them have you recently tried engaging legal services in recent years?
• 2:43
  SPEAKER_00: It is very hard.
• 2:44
  SPEAKER_00: They are very busy, but frequently the demand for legal services already outpaces the supply.
• 2:54
  SPEAKER_00: And it actually has for some time.
• 2:56
  SPEAKER_00: For some years, that's been the case.
• 2:57
  SPEAKER_00: And there's other things.
• 2:59
  SPEAKER_00: So the American Bar Association just published uh an article late last year that talked about over the last 10 years, the number of attorneys in the United States has grown over the entire 10-year span 5%.
• 3:14
  SPEAKER_00: It's like 5.6.
• 3:16
  SPEAKER_00: That is half of a percent every year.
• 3:19
  SPEAKER_00: It is basically flat.
• 3:21
  SPEAKER_00: So is the world different today than it was 10 years ago?
• 3:25
  SPEAKER_00: Yeah, you betcha.
• 3:26
  SPEAKER_00: And it's quite a bit different today than it was 10 years ago.
• 3:28
  SPEAKER_00: And the demand and the underlying things that have changed actually increase the latent demand for legal services.
• 3:36
  SPEAKER_00: And so you have this massive latent demand that is unaddressed.
• 3:40
  SPEAKER_00: Even the kind of those uh able to pay at uh legal prices today, even they have difficulty getting the service that they need.
• 3:50
  SPEAKER_00: So there's way more work already than we have um people and attorneys to do it.
• 3:56
  SPEAKER_00: And so what we really should be talking about is a productivity and extendability, uh extensibility of legal professionals, um, making them be able to do more because we already know that even without AI, we've we've had too few of them uh for the world as it's changed.
• 4:16
  SPEAKER_00: So that's one thing is that uh that's how I would frame this conversation is it's about nobody's job becoming obsolete.
• 4:22
  SPEAKER_00: We have already way too few of them.
• 4:24
  SPEAKER_00: Um AI is not going to change that.
• 4:27
  SPEAKER_00: It's going to make the underlying professionals uh superpower, right?
• 4:31
  SPEAKER_00: It's gonna make them be able to do more things um at a higher quality and at a faster pace.
• 4:36
  SPEAKER_00: So I would say that's kind of um misconception number one.
• 4:42
  SPEAKER_00: I would say, you know, I don't know.
• 4:45
  SPEAKER_02: And that's and I bet a lot of people would be surprised by that because there is this perception of wow, it performs as well as as an attorney, it can create contracts, it can do this, can do that.
• 4:54
  SPEAKER_02: But what you're saying is it's really an accelerator.
• 4:56
  SPEAKER_02: And uh I think I when we first uh uh chatted, I told you a little bit about the work that we used to do in the legal profession.
• 5:04
  SPEAKER_02: I might save that a little bit later, but what but the big thing was we're we're selling efficiency and they were like, oh, really?
• 5:10
  SPEAKER_02: I'm gonna be able to bill fewer hours.
• 5:12
  SPEAKER_02: Yeah, no, thank you.
• 5:13
  SPEAKER_02: Right.
• 5:13
  SPEAKER_00: Yeah, you know, the yeah, go ahead.
• 5:15
  SPEAKER_00: Right, the underlying pricing models, those may adjust.
• 5:18
  SPEAKER_00: But if you're just talking about the legal services output and what demand there is for that output, it is very high.
• 5:25
  SPEAKER_00: It outstrips the supply.
• 5:27
  SPEAKER_00: And there's also kind of a less measured, latent demand of the number of times households come up against civil legal issues and they want help.
• 5:37
  SPEAKER_00: I was just telling a friend recently that a few months ago I wanted legal help on a small real estate issue.
• 5:44
  SPEAKER_00: I called four law firms that were recommended to me online.
• 5:47
  SPEAKER_00: No one got back to me.
• 5:48
  SPEAKER_00: I don't, I don't blame them.
• 5:49
  SPEAKER_00: They just have plenty of work as it is right now.
• 5:52
  SPEAKER_00: And so, yes, it is it does have this underlying question of oh, if the billable hour is the main method of getting paid as an attorney, doesn't a accelerant or kind of supercharging, doesn't it go against that?
• 6:07
  SPEAKER_00: Well, people that you know, the future is dynamic, people adapt.
• 6:09
  SPEAKER_00: And legal professionals will have lots of creative ways to continue to get paid for their services, even if the billable hour has some sort of transformation.
• 6:19
  SPEAKER_02: So, so yeah, and you talk about latent demand, I can I can imagine how that is going to, if if you had more readily available services and you have the capacity that latent demand could be could be met.
• 6:31
  SPEAKER_02: What what else are you seeing in terms of misconceptions and concerns or fears in the in the industry?
• 6:38
  SPEAKER_00: Yeah, I would say just the number the other misconception out there is just how much of business of law there is, right?
• 6:45
  SPEAKER_00: These are businesses.
• 6:46
  SPEAKER_00: A law firm is a business, a legal department inside of a corporation operates like any other department where it has to contribute, it is held accountable in specific ways.
• 6:58
  SPEAKER_00: And so there's just getting deep into the professional work of paralegals and lawyers.
• 7:04
  SPEAKER_00: You realize that yes, a lot of this is getting into the case law, a lot of this is the things that are taught in law school.
• 7:12
  SPEAKER_00: A lot of this is, though, just the administrative burden of like, did so-and-so get the right kind of document?
• 7:19
  SPEAKER_00: We get it back from them, the coordination with clients, the collaboration with outside counsel, making sure all of that happens in a streamlined and easy way that doesn't kind of put a bunch of cognitive load on the folks that are actually going to do the higher order thinking.
• 7:36
  SPEAKER_00: So there's a lot of that.
• 7:37
  SPEAKER_00: There's a lot of coordination costs and delegation costs inside of a legal team.
• 7:44
  SPEAKER_00: And so if you help address some of those things, you actually give them the ability to do, like I said, that knowledge work, that legal thinking uh that they're really paid to do.
• 7:55
  SPEAKER_00: Um so even if going back to the earlier topic that we mentioned, if you if you bill by the hour, right?
• 8:01
  SPEAKER_00: If you if you're a contingency fee attorney, which is like, you know, one in one in ten or something like that, you know, this is less of an issue.
• 8:08
  SPEAKER_00: But if you are if you bill by the hour, um we there's still a way to think about this, is that you have a portion of your day that is non-billable.
• 8:17
  SPEAKER_00: Right.
• 8:18
  SPEAKER_00: And so you need to reduce the amount, yeah.
• 8:21
  SPEAKER_00: You need to reduce those non-billable hours.
• 8:23
  SPEAKER_00: And a lot of that just looks like administrative burden, hopping, you know, getting someone to hop from one system to the next, making sure someone knows that they've been given a task and that they know when it's due.
• 8:35
  SPEAKER_00: Those types of things that are just endemic to uh corporate life, those are a ton of work um and effort done inside of these businesses or legal departments.
• 8:46
  SPEAKER_00: Yeah.
• 8:47
  SPEAKER_02: So, so think about uh, you know, the fact that the the legal industry is very cautious as they should be.
• 8:53
  SPEAKER_02: Um when you look back several years, what was the landscape like?
• 8:58
  SPEAKER_02: Think about adoption.
• 8:59
  SPEAKER_02: When did adoption begin to accelerate?
• 9:01
  SPEAKER_02: When did people start to feel like they could trust these systems or move from older systems into a modern environment where AI could be deployed?
• 9:10
  SPEAKER_02: I I recall I did a session for the uh for the American Bar Association, the future of the legal profession, about 20 years ago.
• 9:16
  SPEAKER_02: And I said, wouldn't it be great if you had you know a device where you could go check all your case matters and all your and people like in in the court and like that's ridiculous, right?
• 9:26
  SPEAKER_02: We were never, you know, how could you do that?
• 9:28
  SPEAKER_02: It's impossible.
• 9:29
  SPEAKER_02: But tell me about that transition and how that adoption accelerated.
• 9:33
  SPEAKER_00: Yeah, it's interesting.
• 9:34
  SPEAKER_00: I've seen now two transitions inside of legal one in the late 2010s of law firms finally, a critical mass of them finally joining kind of cloud-based platforms and you know, having an increasing number of younger generations that wanted access to the case file on their mobile device or whatever it might be.
• 9:53
  SPEAKER_00: So there was that transition.
• 9:56
  SPEAKER_00: And in some ways, I think that transition was probably harder.
• 10:00
  SPEAKER_00: I think in the initial years of AI, there was that skepticism, there was that risk aversion.
• 10:06
  SPEAKER_00: I would say in the last 12 months, we have really kind of crossed into new territory.
• 10:12
  SPEAKER_00: There's been a sea change moment where a lot of them start realizing, no, this actually will help us quite a bit and are very, very interested.
• 10:20
  SPEAKER_00: Yeah, it's 12 to 18 months, where it feels like most of our customers, most of the folks we engage with, right, they are their appetite is wetted and they want to see what these tools can do for them.
• 10:31
  SPEAKER_00: So I feel like the story has been less and less about suspicion.
• 10:34
  SPEAKER_00: Definitely that was the case, say in 2024, 2023, and more interest in what this can do for them.
• 10:42
  SPEAKER_00: I always like to be as a chief product officer, I like to be in a metaphor, more metaphorical terms, like 200 feet in front of our customers in terms of technology and innovation, right?
• 10:55
  SPEAKER_00: Not so far away that it's like, oh, where are they going?
• 11:00
  SPEAKER_00: That you know, they're they're it's become an ivory tower, who knows what's happening.
• 11:04
  SPEAKER_00: And not behind them, where they're like, oh, you know, there's a lot of things we want to be able to do that we can't do.
• 11:08
  SPEAKER_00: And a lot of times I'm now coming to our customers, attorneys, IT directors at law firms, and sometimes I'm just showing them like the bleeding edge of what we're thinking about, what we've prototyped.
• 11:21
  SPEAKER_00: And it is not a like the conversation goes really well, right?
• 11:25
  SPEAKER_00: They get, and I would say that wasn't the case about AI two years ago, but certainly in 2025 and certainly in 2026 now, the openness to it, the quickly accepting it, the kind of they're they're much quicker studies on how this could help them than they used to be in the past.
• 11:43
  SPEAKER_02: Yeah, and and I'm sure the that initial trepidation came from all those well-publicized examples of people going in court with with uh fabricated law, uh case law and fabricated uh testimony and fabricated and all of this.
• 11:58
  SPEAKER_02: And so part of it is the maturity of the industry to understand how do you use this, what's the correct application, what are the guardrails, how do you prevent hallucinations, how do you check for them, and how do you use the the inherent capabilities to do the things that you need to do, like comparing documents or comparing, I know one of the things we talked about were real-time analysis of depositions, right?
• 12:21
  SPEAKER_02: An attorney can't keep everything in their in their minds, but when they're deposing someone, but someone could say something that was inconsistent with other testimonies, could be inconsistent with a police report, could be inconsistent with something else that they said an hour ago.
• 12:33
  SPEAKER_02: Well, it's hard to keep that on that in your mind, but you can do that real time with these tools.
• 12:37
  SPEAKER_02: So talk a little bit about those kinds of things.
• 12:40
  SPEAKER_00: Yeah, I can absolutely talk about those.
• 12:41
  SPEAKER_00: Um so I remember it was probably a couple of years ago now, and maybe a little bit further back, and talking to customers about some AI widgets that we had just added into our platform, right?
• 12:53
  SPEAKER_00: So they're already working in this platform, it's where they do their document review, it's where they delegate things, timekeep, whatever.
• 13:00
  SPEAKER_00: And we offered a few widgets and kind of a library of choices or prompts that someone could say, hey, I want an analysis of this deposition transcript, or I want an analysis of this particular kind of contract, or whatever it might be.
• 13:14
  SPEAKER_00: So there was a whole library of these choices.
• 13:17
  SPEAKER_00: And uh, and yeah, in those early years of trepidation, we had to hunt and find folks that were really taking advantage of it and seemed to be deepening their engagement and repeat users.
• 13:27
  SPEAKER_00: And so we would talk to them, of course.
• 13:28
  SPEAKER_00: And I remember one of those customers was a huge fan of what we offered in terms of analyzing deposition transcripts.
• 13:36
  SPEAKER_00: And uh it would do a really good way of issue spotting things that had happened in the deposition that they didn't even remember, right?
• 13:44
  SPEAKER_00: You're sitting there participating in all of it, and uh you didn't remember that something was said in a certain way.
• 13:52
  SPEAKER_00: And they realize this, and it helps them know that in preparation for other pretrial things or trial.
• 13:58
  SPEAKER_00: And then I remember one customer said to me, Gosh, if only I had this analysis of the deposition during the deposition.
• 14:05
  SPEAKER_00: And uh, you know, we had our product years on and we were like, aha.
• 14:10
  SPEAKER_00: And what that really kicked off was a whole depositions journey for us, where uh the first thing we did was we said, okay, we will live transcribe what is happening during the deposition.
• 14:21
  SPEAKER_00: We will send it through analysis, we'll compare it to the outline or goals that you had during the deposition, and we'll see how you're doing.
• 14:30
  SPEAKER_00: We'll see if the deponent has said something inconsistent with what they said maybe two hours earlier, or something that was inconsistent with an affidavit or a police report or something like that that is already inside the case file.
• 14:43
  SPEAKER_00: And uh, that is now kind of pushed us in a whole direction of a depositions platform where there's AI experiences pre-deposition and preparing for the deposition and putting together that outline during the deposition with this real-time transcription and feedback.
• 14:59
  SPEAKER_00: And then also following the deposition, even in just the administrative things of making sure you have good clips to show during trial, um, and then also analysis of the deposition transcript uh that's certified after the fact.
• 15:10
  SPEAKER_00: So we really took that feedback and lots of other iterations.
• 15:14
  SPEAKER_00: And yes, it was it's a great use case for AI.
• 15:18
  SPEAKER_02: Yeah, and and again, it's one of those things where you know when you understand those workflows and you understand those objectives and what those manual processes would be, then you can you can automate and you can accelerate and provide new insights and new capabilities.
• 15:34
  SPEAKER_02: That's what's so valuable about kind of getting the real-time feedback from your customers and being able to act on that.
• 15:41
  SPEAKER_02: Um, so one of the things that um we talked a little bit about was um, you know, is bullion search and a semantic search and and it's not necessarily an either either or why is uh bullion search still important?
• 15:56
  SPEAKER_02: Um and talk a little bit about uh how that fits in with semantic search and and uh and even retrieval augmented generation.
• 16:03
  SPEAKER_00: Yeah, that that's great questions.
• 16:05
  SPEAKER_00: In legal has something special where if you are responding to, say, let's say you're at attorney general's office and you're responding to a public records request, or there's other document requests, might be part of a subpoena, might be part of a legal hold.
• 16:20
  SPEAKER_00: And then you have to be accountable for how you define the population of documents that were provided to satisfy that request.
• 16:28
  SPEAKER_00: And that that population definition needs to have external transparency that someone could say, okay, well, I see the logic here, and that is a reasonable way to define a set of documents that fulfilled the request.
• 16:42
  SPEAKER_00: So that's that's a reason, a core underlying reason why Boolean search uh can't go away inside legal.
• 16:49
  SPEAKER_00: But you know, what is interesting is the semantic search and RAG has everyone realizing that it's a much better way of finding out information that you actually need to do the task at hand, uh, or you know finding the right document or the hot doc or whatever it might be, then the Boolean search, right?
• 17:09
  SPEAKER_00: You can just take natural language and convert it into uh a rich retrieval across a given matter.
• 17:16
  SPEAKER_00: So uh well, you actually have you have to have both.
• 17:18
  SPEAKER_00: And instead of just saying, okay, because you have this particular thing that might come up in uh with a particular judge evaluating the document population that was provided um for a subpoena or uh a legal hold or whatever it might be, that doesn't mean that has to be the way that you find out information.
• 17:37
  SPEAKER_00: Uh folks that have been in the industry for a long time had a colleague that uh has been doing this for decades, right?
• 17:45
  SPEAKER_00: You internally we started having these light bulbs uh go on and we start realizing, oh, well, just because we need to have that does not mean that it needs to be the primary experience of information gathering inside of a legal organization.
• 17:58
  SPEAKER_00: So yeah, you definitely need both.
• 17:59
  SPEAKER_00: Um I would say there's actually a third one.
• 18:02
  SPEAKER_00: So I would say three lenses.
• 18:04
  SPEAKER_00: One is this retrieval lens that you know go after semantically what I'm after.
• 18:08
  SPEAKER_00: Two is I need to define in in Boolean terms that other people would recognize that we talked about.
• 18:14
  SPEAKER_00: And then the third one is I have some underlying system where my matter, where my matters reside, and I want to use the attributes of that system to find out information.
• 18:24
  SPEAKER_00: So I think there's actually three lenses that a legal professional needs in order to gather information.
• 18:29
  SPEAKER_00: And they they want a combination or an isolation of each one of those three, and that's what a great AI experience needs to support.
• 18:36
  SPEAKER_02: Yeah, yeah.
• 18:37
  SPEAKER_02: And I think that last uh parameter is really attribute-based.
• 18:41
  SPEAKER_02: Exactly.
• 18:42
  SPEAKER_02: And that's part of retrieval augmented generation, right?
• 18:44
  SPEAKER_00: Because you do that's true.
• 18:45
  SPEAKER_02: That information protector.
• 18:47
  SPEAKER_02: I was doing work in uh discovery analysis for a case that had enormous numbers of documents at one point.
• 18:54
  SPEAKER_02: Understanding the topics, understanding the objectives, understanding the details of those documents and being able to search, you know, full text search or bullion search, it was really, really complex, right?
• 19:07
  SPEAKER_02: And it really required a lot of thinking through in advance what those topics might look like, right?
• 19:14
  SPEAKER_02: So you really had to do some information architecture uh in order to do it correctly.
• 19:17
  SPEAKER_02: So just talk a little bit about what needs to be in place.
• 19:20
  SPEAKER_02: So, you know, when you think about AI readiness, a lot of organizations are want AI, they want to be able to deploy it, but they don't have the foundations.
• 19:28
  SPEAKER_02: So, what do you see?
• 19:29
  SPEAKER_02: What do you think an organization needs to do to make AI work in a real environment?
• 19:34
  SPEAKER_02: And where do you see teams struggling, whether it's content structure, systems of integration, data acts uh access, or workflow maturity?
• 19:43
  SPEAKER_00: Yeah, uh so much to say on this topic.
• 19:45
  SPEAKER_00: So to piggyback off of what you were talking about before, you need tools and uh an underlying attribution system.
• 19:53
  SPEAKER_00: And maybe it's not the place where, you know, in a in a gentic AI world, maybe an agent is interfacing with these systems, but the underlying system still needs to have the ability to attribute things.
• 20:05
  SPEAKER_00: Um and it still needs the thing, you know, you should be able to say, okay, this object or artifact has these characteristics and they, you know, have their own, you know, their own fields, they have their own tags, they have uh this is very common, and in some ways, legal thought about this, legal industry thought about this long before uh there were L or any LLMs.
• 20:27
  SPEAKER_00: When you upload a document into a lot of legal systems, you were asked to give it a document profile and uh various attributes of that document before it's loaded into the system.
• 20:39
  SPEAKER_00: And that is to say, okay, when we're doing a big search or a big e-discovery or something like that, we can use that profile that was filled out about that document in advance.
• 20:50
  SPEAKER_00: I would say look at the various kind of core objects inside of the colour, and that would be a legal matter, it could be a contract or or some sort of deal.
• 21:02
  SPEAKER_00: It could be the documents themselves.
• 21:05
  SPEAKER_00: It could be it could be a dep, it could be a deposition, it could be um the you know the the client and all of the attributes that they have.
• 21:16
  SPEAKER_00: It could be the outside council guidelines, right?
• 21:19
  SPEAKER_00: There's so many things that actually need uh they that need their own set of attributes.
• 21:26
  SPEAKER_00: And so your system needs to be able to support that, even if you don't, right, even if you just sit above a kind of an agentic layer um and don't engage with those underlying user interfaces, you still need something that handles that.
• 21:39
  SPEAKER_02: Yep.
• 21:40
  SPEAKER_02: I call we call it the isness and aboutness, right?
• 21:42
  SPEAKER_02: What is this thing?
• 21:43
  SPEAKER_02: It's a contract.
• 21:44
  SPEAKER_02: What is it about?
• 21:45
  SPEAKER_02: It's about these parties, right?
• 21:46
  SPEAKER_02: It's about it's about this contract type, this transaction, whatever it might be.
• 21:51
  SPEAKER_02: And all of that is core information architecture.
• 21:54
  SPEAKER_02: And it also brings up the idea of you know, before large language models, I mean, AI didn't you know arise in 2023.
• 22:00
  SPEAKER_02: A lot of people think it did.
• 22:02
  SPEAKER_02: It's been around for a long time.
• 22:04
  SPEAKER_02: Right?
• 22:04
  SPEAKER_02: We used to call it text analytics.
• 22:06
  SPEAKER_00: We used to call it presentation, right?
• 22:08
  SPEAKER_02: We used to call it all these things.
• 22:10
  SPEAKER_02: We've been working with machine learning and documents and content and data for you know 25 years, right?
• 22:16
  SPEAKER_02: 30 years.
• 22:17
  SPEAKER_02: And so it's been around for decades and decades.
• 22:20
  SPEAKER_02: So uh it's interesting when you when you think about that, and then that is what is required still today, uh, is how that uh content and that that content structure and that system integration and those attributes do have to be defined.
• 22:34
  SPEAKER_02: They have to be defined for the case itself, right?
• 22:37
  SPEAKER_02: Because it's gonna have unique characteristics that may be different than your organizational taxonomies and architecture.
• 22:45
  SPEAKER_02: But the case itself is gonna have its own issues, its own attributes, its own is-ness and aboutness, right?
• 22:50
  SPEAKER_02: Yes, that's exactly right.
• 22:52
  SPEAKER_00: And we've seen this happen in real time with our customers figuring out that the is-ness and aboutness has a lot of value, even if they don't actually see it.
• 23:01
  SPEAKER_00: And I'll give you an example.
• 23:02
  SPEAKER_00: So uh we have an agent in our platform, we call it Lois, legal, uh, legal operating intelligence system.
• 23:10
  SPEAKER_00: Okay.
• 23:11
  SPEAKER_00: And customers go in there and they ask questions about their specific case or matter.
• 23:17
  SPEAKER_00: And there's a significant portion of those questions that are things like what are the insurance policy limits or whatever.
• 23:25
  SPEAKER_00: I'm just uh something off the top of my head.
• 23:28
  SPEAKER_00: Well, there is an application interface that they could find out what were the policy limits for this particular matter, but they instead decided to ask.
• 23:38
  SPEAKER_00: Because we introduce a new interface that is kind of a single surface interface where they can ask and get answers to these questions.
• 23:47
  SPEAKER_00: It does not mean that you you don't need a place where the actual policy limit is stored for quick recall, right?
• 23:54
  SPEAKER_00: And so we've we've seen this play out exactly as you're describing.
• 23:59
  SPEAKER_02: So it's interesting when you think about uh, and and there are a lot of ways of doing that, right?
• 24:04
  SPEAKER_02: You can automate many of those pieces, you can automate classification, but you still need that reference architecture, right?
• 24:10
  SPEAKER_02: You still need to build the values.
• 24:12
  SPEAKER_02: What are the fields, what are the entities, what are the vocabularies, and then you can run these things against uh uh your corpus and be able to do the classification.
• 24:21
  SPEAKER_02: So that is still you know content operations, it is still content governance, right?
• 24:25
  SPEAKER_02: It is still architecture, it's still building ontologies, right?
• 24:29
  SPEAKER_02: All of those are still so critical.
• 24:31
  SPEAKER_02: We should talk about how your customers do that now and where we can help you automate more of that.
• 24:35
  SPEAKER_02: So at some point, but uh a top, another topic for another conversation.
• 24:39
  SPEAKER_02: What do you see is the biggest impact of AI over the coming years?
• 24:44
  SPEAKER_02: Where do you see this going for organizations?
• 24:48
  SPEAKER_00: Interface is the biggest impact.
• 24:50
  SPEAKER_00: Interface is by sh by far the biggest impact.
• 24:53
  SPEAKER_00: That is the big unlock.
• 24:54
  SPEAKER_00: Uh, like you said, a lot of the things, you know, the forces of gravity still exist.
• 24:59
  SPEAKER_00: You still need the isness and aboutness, but the interface changes.
• 25:03
  SPEAKER_00: And uh a big challenge as someone who, you know, eight years ago was helping law firms get in from like a legacy on-prem system to a cloud-based system, the adoption curve of like, well, this is now where I go to do XYZ, or I now have to know my click path, or I now have to know how to navigate this.
• 25:22
  SPEAKER_00: And that that huge switch carried a big adoption curve and frustration and all of that, right?
• 25:29
  SPEAKER_00: The the interface is the unlock, the ability to just ask a question, the ability to make an assignment, the ability to track a time entry, the ability to set an appointment or set a deadline and know what are the statutory deadlines that are tied to that anchor date, right?
• 25:48
  SPEAKER_00: Those are all things that now someone that doesn't know the click path um in a particular system that they can find out and do their job in a really straightforward way.
• 25:59
  SPEAKER_00: So for sure it is the interface.
• 26:01
  SPEAKER_00: I mean, you think about in and if you think about folks, a lot of um a lot of these like I had a neighbor who practiced law well into his 90s.
• 26:12
  SPEAKER_00: Um, and that interface of being able to just ask a question, that's huge.
• 26:17
  SPEAKER_00: That's huge.
• 26:18
  SPEAKER_00: The ability to just type out what you want to have happen.
• 26:21
  SPEAKER_00: So the interface is by far the biggest change.
• 26:24
  SPEAKER_00: Um, I think everything else really is is well the number two biggest change is is is a far too.
• 26:30
  SPEAKER_00: It's a distant second because the interface is such a great way to work.
• 26:34
  SPEAKER_00: And I find myself, if I find a system that I can't interact with, right?
• 26:39
  SPEAKER_00: I deal with various product and tech systems in my day-to-day job.
• 26:43
  SPEAKER_00: And if I find one that like I don't have that interface, now I get annoyed because I want to work with technology in that way now.
• 26:49
  SPEAKER_02: Absolutely.
• 26:50
  SPEAKER_02: And that conversational interaction is really going to be, you know, how things are are evolving and will continue to evolve.
• 26:57
  SPEAKER_02: And we'll see the ROI from all of the efficiencies that we're gaining and the new capabilities, and as you say, meeting that latent um uh demand for these uh for legal services.
• 27:09
  SPEAKER_02: Well, this has been tremendous.
• 27:10
  SPEAKER_02: Thank you so much for your time.
• 27:12
  SPEAKER_02: I've really enjoyed the conversation.
• 27:14
  SPEAKER_02: Any any final thoughts before we uh finish up?
• 27:16
  SPEAKER_00: Uh, this has been great, Seth.
• 27:18
  SPEAKER_00: Really appreciate your time.
• 27:18
  SPEAKER_00: I've enjoyed it.
• 27:20
  SPEAKER_00: Absolutely.
• 27:20
  SPEAKER_02: And thank you to our audience.
• 27:23
  SPEAKER_02: Thanks for listening to uh the Early AI podcast.
• 27:26
  SPEAKER_02: Please subscribe, share it with colleagues, and we will see you next time.
• 27:30
  SPEAKER_02: Thank you.
• 27:32
  SPEAKER_01: Thank you for joining us on another deep dive into AI innovation.
• 27:36
  SPEAKER_01: Tune in next time when we introduce another industry expert and discuss how to maximize AI in your world.
• 27:42
  SPEAKER_01: The Early AI Podcast is sponsored by Early Information Science and Vector.
• 27:47
  SPEAKER_01: To learn more, visit early.com.
• 27:50
  SPEAKER_01: That's E A R L E Y dot com.
• 27:53
  SPEAKER_01: Thank you for tuning in.