1 00:00:00,151 --> 00:00:01,772 Kyle Poe, how are you this afternoon? 2 00:00:01,772 --> 00:00:05,163 I'm doing awesome. 3 00:00:05,163 --> 00:00:08,225 This is actually the third podcast I've recorded today. 4 00:00:08,225 --> 00:00:10,916 um Yeah, today's a busy day. 5 00:00:10,916 --> 00:00:18,871 Normally I do like one a week, but we had a ton of travel in the late fall, so we burned through our backlog. 6 00:00:18,871 --> 00:00:24,244 Normally we keep a big backlog, so I'm rebuilding the pipeline again, so to speak. 7 00:00:24,244 --> 00:00:25,335 Well, it's great to have you on. 8 00:00:25,335 --> 00:00:32,040 You and I got to know each other before, back when you were in the, still on the law firm side, you've now moved to the vendor side. 9 00:00:32,040 --> 00:00:34,842 And it's great to finally have you on the show. 10 00:00:34,842 --> 00:00:35,962 We got a great agenda. 11 00:00:35,962 --> 00:00:42,606 Before we jump in, why don't you just do a quick introduction and tell us who you are, what you do and where you do it. 12 00:00:42,984 --> 00:00:43,754 Sure. 13 00:00:43,754 --> 00:00:46,886 So Kyle Poe, VP of legal innovation and strategy here at Legora. 14 00:00:46,886 --> 00:00:53,589 So former practicing attorney, I practiced law for 14 years, most of which at Morgan Lewis, where I was an associate and then a partner there. 15 00:00:53,589 --> 00:01:01,192 uh My practice was primarily mass tort and product liability litigation, but I also dabbled in labor and employment as well as just other commercial litigation. 16 00:01:01,212 --> 00:01:05,344 Really early in my career, I got involved in the tech side of legal practice. 17 00:01:05,344 --> 00:01:09,588 So building custom technology solutions internally at Morgan Lewis. 18 00:01:09,588 --> 00:01:15,068 which led me ultimately to using AI at the firm and using it for my own practice. 19 00:01:15,068 --> 00:01:22,048 I got to see firsthand how powerful AI was and that gave me the conviction that this was really going to be a game changer. 20 00:01:22,048 --> 00:01:31,348 I had been doing legal tech and building software in the firm for a decade and nothing before that was as powerful or as compelling as AI had been. 21 00:01:31,768 --> 00:01:38,588 I wanted to be part of this wave and so I decided to leave at that time and go on to the vendor side as you said. 22 00:01:38,866 --> 00:01:40,238 led me to here at Legora. 23 00:01:40,238 --> 00:01:48,384 And what I do here is I work with customers, help them understand the technology as well as adapt it for their own practices as well as their own business. 24 00:01:48,384 --> 00:01:55,929 So really thinking about the future of legal practice and how this is changing not only what they're doing on a day-to-day basis, but also kind of the next level in terms of how 25 00:01:55,929 --> 00:02:00,232 they're reshaping their pricing, their packaging, their delivery of legal services using AI. 26 00:02:00,648 --> 00:02:01,669 Yeah, you know, it's interesting. 27 00:02:01,669 --> 00:02:06,393 You don't see many partners that make the transition into the tech side of the business. 28 00:02:06,393 --> 00:02:07,565 Like what drove that? 29 00:02:07,565 --> 00:02:08,937 Was it just your passion? 30 00:02:08,937 --> 00:02:14,477 Yeah, I I had seen, I mean, there have been a couple of us who have made that jump. 31 00:02:14,617 --> 00:02:24,337 I think it's, it is unprecedented, but I think this is we sort of live in unprecedented times, you know, is AI, there's nothing like AI in terms of the breadth of experience, you 32 00:02:24,337 --> 00:02:28,437 know, I had been doing innovation internally, as I said at the firm. 33 00:02:28,477 --> 00:02:31,717 And, you know, I did it originally for my own master of practice. 34 00:02:31,717 --> 00:02:35,497 And then I got a reputation as being like the tech enabled lawyer. 35 00:02:35,497 --> 00:02:38,121 And so people would come to me, you know, I would speak to 36 00:02:38,121 --> 00:02:40,141 you know, partners throughout the firm. 37 00:02:40,141 --> 00:02:42,101 And they would ask like, hey, can you help? 38 00:02:42,101 --> 00:02:48,461 And I built up a team of data analysts and engineers and developers, a team of about 10 by the time I left. 39 00:02:48,461 --> 00:02:50,861 And so I had resources to deploy. 40 00:02:51,101 --> 00:02:55,661 But most of the time when I would talk to partners and I would ask them, okay, you know, what do you do on a day-to-day basis? 41 00:02:55,661 --> 00:02:57,641 You know, how's your business structured? 42 00:02:57,641 --> 00:02:58,541 What are your associates doing? 43 00:02:58,541 --> 00:02:59,901 Your mid-levels doing? 44 00:03:00,041 --> 00:03:01,461 What are your pain points? 45 00:03:01,681 --> 00:03:05,673 Most of the time, like 90 % of the time, I would come back to them and say, you know, 46 00:03:05,673 --> 00:03:18,033 whatever I'm building, which is a lot of databases, matter management, statistical analytics, most of those approaches, kind pre-gen AI technologies, weren't really adept at 47 00:03:18,033 --> 00:03:21,093 helping the vast majority of the partners that I would talk to. 48 00:03:21,093 --> 00:03:29,293 And so for the most of the time, I would just tell them, what I have and what I'm able to do isn't really going to make a huge impact for your practice. 49 00:03:29,293 --> 00:03:34,633 Basically, before gen AI, for you to have a good use case using pre-gen AI technologies, 50 00:03:34,633 --> 00:03:39,693 You need to have a practice that had a lot of repeatability to it, that had a lot of scale to it. 51 00:03:39,693 --> 00:03:48,313 Often you need to find a practice that already sort of had the right business model in place, say like a fixed fee or alternative fees or contingency fees, for it to really make 52 00:03:48,313 --> 00:03:50,863 sense to deploy pre-gen AI technologies. 53 00:03:50,863 --> 00:03:54,403 And so when GenAI came out and I started using it, it was just so clear. 54 00:03:54,403 --> 00:03:56,443 mean, look, computers learn to speak English. 55 00:03:56,443 --> 00:04:00,663 They learn to speak natural language for the first time with GenAI and LLMs. 56 00:04:00,803 --> 00:04:03,463 the basically 90, there's 90 % of 57 00:04:03,463 --> 00:04:11,288 partners I spoke to who I could no longer help or I couldn't help before, all of a sudden I was in position to go back to them and tell them the time is now. 58 00:04:11,288 --> 00:04:13,610 The technology has advanced to that point. 59 00:04:13,610 --> 00:04:15,730 And so I was ready to go all in. 60 00:04:15,730 --> 00:04:22,198 think others, this is back in 2023, 2024 timeframe, not everyone was so sold on the technology. 61 00:04:22,198 --> 00:04:28,670 think there was a, especially at that time, the initial knee jerk reaction of a lot of clients was like, don't use gen AI. 62 00:04:28,925 --> 00:04:32,016 There's a lot of fear, I think, early on, whereas I was leaning in. 63 00:04:32,016 --> 00:04:40,910 And so it just became clear that if I wanted to have the impact, a broad impact, that the moment demanded that it would be outside the firm at that point. 64 00:04:41,062 --> 00:04:41,612 Interesting. 65 00:04:41,612 --> 00:04:52,237 Yeah, and your days inside the firm, you had a big part in the ExtraNet that Morgan Lewis kind of built their own custom ExtraNet platform on SharePoint. 66 00:04:52,238 --> 00:04:52,778 Correct. 67 00:04:52,778 --> 00:04:53,138 Yeah. 68 00:04:53,138 --> 00:04:55,839 Was that SharePoint on-prem or SharePoint online? 69 00:04:57,160 --> 00:04:57,694 Okay. 70 00:04:57,694 --> 00:05:00,040 I think we eventually moved to SharePoint Online. 71 00:05:00,040 --> 00:05:00,800 Yeah. 72 00:05:00,800 --> 00:05:05,060 So, and that's kind of how you and I got connected was your experience around that. 73 00:05:05,060 --> 00:05:13,560 I've spent the last 17 years in the legal collaboration space started in 2008 with a company called Acrowire. 74 00:05:13,560 --> 00:05:21,660 And we did a lot of time and materials consulting, building custom solutions like the one that you guys built in-house. 75 00:05:21,660 --> 00:05:29,376 Most firms don't have those have the chops, you know, or the personnel to deliver on an engagement like that. 76 00:05:29,539 --> 00:05:40,814 Morgan Lewis being a big firm obviously did, but yeah, you learned a ton in that process and then your current role, you've kind of translated that and helped them with their 77 00:05:40,814 --> 00:05:42,445 portal product, correct? 78 00:05:42,696 --> 00:05:43,388 That's right, yeah. 79 00:05:43,388 --> 00:05:51,011 Tell us a little bit about kind of what the thinking was with Portal and where you see it going. 80 00:05:51,389 --> 00:06:00,849 Yeah, so the work was, you know, from the early days was envisioned to be a collaborative AI workspace for lawyers where you could collaborate initially internally within a firm. 81 00:06:00,849 --> 00:06:06,089 So you can share it between different teams who work in a firm, share content more broadly. 82 00:06:06,089 --> 00:06:16,189 As we were working with firms to develop work, you know, they would create workflows that came out earlier this year, the ability to kind of automate multiple multi-step workflows 83 00:06:16,189 --> 00:06:20,085 using AI and these fully agentic experiences. 84 00:06:20,085 --> 00:06:28,691 As we were building that out for uh our clients, for our customers, they uh began asking, this is really cool, this is really awesome that they have these experiences. 85 00:06:28,691 --> 00:06:31,974 And they were wondering if they could share that with their clients directly. 86 00:06:31,974 --> 00:06:35,216 broadly speaking, the way we see it is a couple of fronts. 87 00:06:35,216 --> 00:06:39,339 One is that AI is changing the speed of delivery. 88 00:06:39,339 --> 00:06:45,595 So the responsiveness that a firm can be now is far greater than it ever could have been before. 89 00:06:45,595 --> 00:06:57,022 the time of performing certain types of legal tasks, particularly things like first pass reviews or things like due diligence, these kinds of tasks are being accelerated greatly. 90 00:06:57,022 --> 00:07:05,216 And as you're able to complete these tasks a lot faster, clients expect faster, more responsive delivery of those services. 91 00:07:05,216 --> 00:07:09,609 And something like email, which is really what most of us are competing against, right? 92 00:07:09,609 --> 00:07:13,413 So email is by far the dominant way. 93 00:07:13,413 --> 00:07:16,026 that lawyers and clients collaborate today. 94 00:07:16,026 --> 00:07:22,561 Like that's the big competitor out there is uh just email and like kind of the status quo. 95 00:07:22,561 --> 00:07:28,725 And so that was maybe a suitable format when it took a month to do a due diligence. 96 00:07:28,725 --> 00:07:38,946 know, having the results come to you in a static Excel spreadsheet maybe was a suitably slow delivery mechanism and suitably, you know, advanced mechanism. 97 00:07:38,946 --> 00:07:43,446 for that kind of legacy work, work performed in a legacy manner. 98 00:07:43,546 --> 00:07:54,726 But once you can complete that work within a matter of days or within a day now, having it be performed by an AI and then double checked by attorneys, clients expect a quicker, a 99 00:07:54,726 --> 00:07:58,066 more enhanced service and delivery of that work product. 100 00:07:58,066 --> 00:08:04,666 And so now you're delivering the work product directly through a portal where they can then interact with the work product. 101 00:08:04,666 --> 00:08:08,226 So they can then use AI to then ask questions of that work product. 102 00:08:08,226 --> 00:08:11,378 And so it just creates this overall better experience for the customer. 103 00:08:11,624 --> 00:08:19,270 Yeah, and you and I had a conversation about kind of where, where Lagora portal fits into the bigger picture. 104 00:08:19,270 --> 00:08:28,525 And I think it, the, the capabilities that Lagora portal provides are a very narrow, but important slice of the big picture. 105 00:08:28,525 --> 00:08:30,256 I shared a graphic with you. 106 00:08:30,256 --> 00:08:31,987 I can share it quickly on the screen. 107 00:08:31,987 --> 00:08:33,528 Most people listen to this podcast. 108 00:08:33,528 --> 00:08:36,220 They don't actually, I might not be able to let's see here. 109 00:08:36,343 --> 00:08:49,654 just to kind of put in context where I see legora portal playing, which is really in that kind of practice space where, you know, intake document, Q and a playbooks, co-authoring, 110 00:08:49,735 --> 00:08:55,240 drafting, shared tabular review, all those sorts of things are very practice of law specific. 111 00:08:55,240 --> 00:09:00,445 And I see that as a necessary, but still fairly narrow slice. 112 00:09:00,445 --> 00:09:02,566 I expect that slice to get bigger. 113 00:09:02,619 --> 00:09:12,227 Over time as AI's capabilities grow and more work from the practice gets fed, becomes tech enabled. 114 00:09:12,227 --> 00:09:23,255 Um, but you know, when we were at a TLTF summit, you guys announced portal, like, I don't know, two or three days and everybody's like, Hey, are you worried about, you know, ex 115 00:09:23,255 --> 00:09:24,236 your extra net product? 116 00:09:24,236 --> 00:09:27,318 And I was like, no, these things are really complimentary. 117 00:09:27,359 --> 00:09:32,082 If anything, and I wrote an article about it on LinkedIn, I'm actually excited. 118 00:09:32,146 --> 00:09:45,747 that you guys and then Harvey later announced shared spaces, I'm excited that there's real money being put into this collaboration space because it's been so badly neglected. 119 00:09:45,747 --> 00:09:50,792 And you know this because you helped build a tool to facilitate it. 120 00:09:50,792 --> 00:09:59,595 It's been so badly neglected that there's very little overlap with what we do, really very little to anything in the practice of law space. 121 00:09:59,701 --> 00:10:11,628 we're more focused on the broader relationship, know, like open and closed matters, historical, you know, legal spend, attorney profile information that are billing time to 122 00:10:11,628 --> 00:10:18,081 the matters, the legal team, all of that sort of stuff, model document sharing, more kind of business of law stuff. 123 00:10:18,201 --> 00:10:25,704 So yeah, I was actually excited, but it really, the reason I wrote that article, I was shocked on how many people 124 00:10:25,750 --> 00:10:29,744 got confused by thinking that what you were doing competed with us. 125 00:10:29,744 --> 00:10:34,067 And these were by, I had a couple of former high Q guys asked me this. 126 00:10:34,067 --> 00:10:37,760 And I'm like, dude, you know what, you understand the landscape. 127 00:10:37,760 --> 00:10:40,191 I'm surprised that you think so. 128 00:10:40,291 --> 00:10:49,028 Anyway, we're glad to see you guys come in and actually in Harvey too, and put some investment here because it's been a badly neglected space. 129 00:10:49,028 --> 00:10:54,383 Yeah, I think that, you know, especially as somebody who, I mean, I first got involved in extranets really early in my career. 130 00:10:54,383 --> 00:11:01,747 And, you know, I certainly I had an impact where, you I own practice using them and delivering, you know, providing client reporting through these, through these portals. 131 00:11:01,747 --> 00:11:05,599 So the clients can understand what we were doing, like with the results that we were getting for our clients. 132 00:11:05,599 --> 00:11:09,071 So like communicating to them the substantive results of our practice. 133 00:11:09,071 --> 00:11:12,346 And certainly, you know, I think that the 134 00:11:12,346 --> 00:11:18,990 The adoption though of portals or extranets as know, oftentimes they're called has been fairly limited, right? 135 00:11:18,990 --> 00:11:25,343 It's been, you I mean, basically a little bit more advanced than a secure file transfer in a lot of instances. 136 00:11:25,343 --> 00:11:36,299 And so the potential, I think a lot of people have seen for a long time, myself included, like we saw robust potential in these client collaboration portals and enhancing the 137 00:11:36,299 --> 00:11:39,472 collaboration between uh law firms and their in-house counsel. 138 00:11:39,472 --> 00:11:41,543 But it's always, it's never lived up. 139 00:11:41,543 --> 00:11:42,893 to that potential. 140 00:11:42,893 --> 00:11:47,855 And I think that, even the things I did with Laura Lewis, just, we barely were scratching the surface. 141 00:11:47,855 --> 00:11:54,528 So I think that when I look at that pie chart you presented, I think it's probably more or less accurate as to where portals are today. 142 00:11:54,528 --> 00:12:03,621 I think that you're absolutely right that like portals today are not delivering, people really aren't delivering a lot of legal services, a lot of final work products. 143 00:12:03,621 --> 00:12:06,411 Certainly they're not creating new service lines, right? 144 00:12:06,411 --> 00:12:08,142 Like self-service legal service. 145 00:12:08,142 --> 00:12:08,690 uh 146 00:12:08,690 --> 00:12:11,570 service delivery through portal today. 147 00:12:11,570 --> 00:12:15,970 Very little of that's being done, particularly in the AMLOS space. 148 00:12:16,330 --> 00:12:17,990 So there's so much more upside. 149 00:12:17,990 --> 00:12:19,310 I think we're going to see a couple of things. 150 00:12:19,310 --> 00:12:21,890 One is you're going to see that pie grow. 151 00:12:21,890 --> 00:12:28,690 The demand for portal services, both on the business law and the practice of law side, is going to skyrocket. 152 00:12:28,690 --> 00:12:30,790 Again, just because we're so early days. 153 00:12:30,790 --> 00:12:36,658 And then I think also, just because practices of law has been such a small portion of what that is historically, 154 00:12:36,658 --> 00:12:37,718 if that's going to grow. 155 00:12:37,718 --> 00:12:39,778 And I think AI is going be the thing that drives that, right? 156 00:12:39,778 --> 00:12:44,938 Like I said earlier, it's going to be one, just you expect to see better experiences. 157 00:12:44,938 --> 00:12:54,489 know, like email is no longer suitable in an age of AI, clients know that their firms are using AI throughout the process to accelerate delivery. 158 00:12:54,489 --> 00:13:00,389 And then as firms are going the next, I think the next step though, is as firms are productizing legal services, right? 159 00:13:00,389 --> 00:13:01,801 So this is kind of the big... 160 00:13:01,801 --> 00:13:07,886 conversation I'm having the most with like management committees and managing partners or firms is uh what's on the horizon. 161 00:13:07,886 --> 00:13:17,573 You the idea that AI is going to disrupt certain things, you know, that it's going to change, you you won't be able to build like, know, when you can take that due diligence 162 00:13:17,573 --> 00:13:22,938 from 30 hours to like, to, you know, one or two, you can't charge 30 hours anymore, right? 163 00:13:22,938 --> 00:13:26,904 Like you can maybe charge two or better yet you reprice that work, right? 164 00:13:26,904 --> 00:13:30,867 You say, Hey, it's no longer be doing, you know, the value of the work. 165 00:13:31,027 --> 00:13:39,134 Maybe that correlates to the 30 hours, but now that it's two hours of attorney time, the value to the client is still substantial for that work. 166 00:13:39,134 --> 00:13:41,226 And it's not really reflected by those two hours. 167 00:13:41,226 --> 00:13:45,218 The metric has gone out of sync for the value that's being delivered to the client. 168 00:13:45,218 --> 00:13:46,909 And so you then want to reprice that work. 169 00:13:46,909 --> 00:13:51,373 Also, it doesn't account for the fact that you had to procure legal technology. 170 00:13:51,373 --> 00:13:56,327 It didn't account for the time and materials to build out the AI workflows that you're using. 171 00:13:56,327 --> 00:13:59,020 And so you need to capture that value somehow. 172 00:13:59,020 --> 00:14:07,811 And so if you reprice that work on a fixed fee basis or on a subscription fee basis, you're then able to, you you productize these things. 173 00:14:07,811 --> 00:14:17,871 And so as firms are seeing kind of what's on the horizon, what's really inspiring to me is that a lot of managing companies are asking like, how can we self disrupt, right? 174 00:14:17,871 --> 00:14:26,391 Like they don't want to be caught off guard and wake up one day when the phone stops ringing and realize the phone has stopped ringing or wake up one day and realize that 175 00:14:26,391 --> 00:14:27,435 they've been undercut. 176 00:14:27,435 --> 00:14:33,915 you know, on an RFP because their competitors are offering stuff on a fixed basis and they haven't done the groundwork, right? 177 00:14:33,915 --> 00:14:38,095 They haven't learned the technology, they haven't built out their workflows, they haven't productized their legal services. 178 00:14:38,095 --> 00:14:40,175 And so they're like caught off guard. 179 00:14:40,175 --> 00:14:42,275 And so a lot of them are being proactive. 180 00:14:42,275 --> 00:14:51,095 Now, obviously, I'm talking to more proactive firms, you know, the ones who have their heads completely in the sand about this, you know, probably aren't taking my calls or 181 00:14:51,095 --> 00:14:53,055 aren't taking the phone to call me. 182 00:14:53,055 --> 00:14:56,395 So I am talking to more forward thinking firms. 183 00:14:56,395 --> 00:15:03,035 But a lot of them are going that next step and really asking from the top down, asking the management committee is thinking about this. 184 00:15:03,035 --> 00:15:11,635 And then they're asking their partners, like throughout the partnership, kind of on a partner by partner basis, what are you doing to rethink pricing, packaging, and delivery 185 00:15:11,635 --> 00:15:13,706 of legal services in light of AI? 186 00:15:13,706 --> 00:15:20,846 One thing that one firm that I was talking to did that was particularly interesting is asking that as part of their year-end partner evaluations. 187 00:15:20,846 --> 00:15:24,906 So at the end of every year, you have to do a self-evaluation as a partner or a questionnaire. 188 00:15:25,002 --> 00:15:29,393 And they'll ask questions like, what have you done for continuity, business continuity? 189 00:15:29,393 --> 00:15:37,298 Like what have you done this year to ensure that if you were to leave the firm or if you were to the lottery or something that retire, that other people will be able to continue 190 00:15:37,298 --> 00:15:39,018 to the torch forward. 191 00:15:39,018 --> 00:15:40,939 So that's kind of a typical question they would ask. 192 00:15:40,939 --> 00:15:43,781 And now they're asking, what have you done with AI, right? 193 00:15:43,781 --> 00:15:46,553 In order to rethink legal services pricing. 194 00:15:46,553 --> 00:15:49,134 And they're getting ahead of things and re-productizing that. 195 00:15:49,134 --> 00:15:51,540 And then the way, once you productize things, 196 00:15:51,540 --> 00:15:58,563 Meaning you take your knowledge and experience of your firm and you embed it into reusable workflows that are AI powered. 197 00:15:58,563 --> 00:16:01,204 You then have to deliver those some power. 198 00:16:01,204 --> 00:16:08,986 And you could just have it be, one option would be to have it just be something that's available to the associate and the associate pulls the lever and then emails the work 199 00:16:08,986 --> 00:16:09,666 product. 200 00:16:09,666 --> 00:16:14,868 But clients know that you're just pulling a lever and they don't want to really, it's not like, it's not a great experience. 201 00:16:14,868 --> 00:16:18,203 And there's just so much more we can do in terms of like interactive experiences. 202 00:16:18,203 --> 00:16:27,348 the ability to ask follow-up questions to able to provide due diligence in this tabular review, which you can do with Legora, which is this interactive spreadsheet-like 203 00:16:27,348 --> 00:16:32,251 experience where you can go into each cell and you can see what the AI's reasoning was. 204 00:16:32,251 --> 00:16:38,003 You can see an audit log of when the AI generated the answer, the associate who signed off on that answer. 205 00:16:38,003 --> 00:16:44,465 To be able to have that level of transparency into the work product is something that we weren't able to have before. 206 00:16:44,465 --> 00:16:52,630 And so yeah, as they're rethinking productizing, a lot of them are looking at the portal as a way to deliver these new legal services directly to their clients. 207 00:16:52,630 --> 00:17:03,922 Yeah, and you and I talked a little bit about like, why is it that such little sophistication has made its way into these client facing portals? 208 00:17:03,922 --> 00:17:07,025 And I think it's a combination of two things. 209 00:17:07,025 --> 00:17:09,127 I'm curious your take on this. 210 00:17:09,127 --> 00:17:14,271 I think that just inherently, law firms are reluctant to share. 211 00:17:14,351 --> 00:17:15,953 They don't want to overshare. 212 00:17:15,953 --> 00:17:17,942 And that's a very 213 00:17:18,000 --> 00:17:19,501 lawyer mindset, right? 214 00:17:19,501 --> 00:17:23,693 And it's a good one from a legal perspective, right? 215 00:17:24,173 --> 00:17:28,156 Additional information that's not essential can create liability, right? 216 00:17:28,156 --> 00:17:29,257 In a legal context. 217 00:17:29,257 --> 00:17:36,402 And so law firms have a, they have a natural stance of being conservative on what they share. 218 00:17:36,402 --> 00:17:39,464 And then clients haven't really demanded it. 219 00:17:39,464 --> 00:17:43,806 Like, you know, we compete primarily with high Q and the number of 220 00:17:43,904 --> 00:17:55,260 Like basic file sharing sites that we see, it's probably four or five to one in terms of anybody doing anything sophisticated with it. 221 00:17:55,260 --> 00:18:00,843 And you know, their architecture is very much kind of isolated and it made a lot of sense. 222 00:18:00,843 --> 00:18:09,368 Like when Haiki first came out, the cloud was big and scary to law firms, you know, so you really, they had to do a lot of work to ring fence and isolate. 223 00:18:09,368 --> 00:18:12,364 And that is really what allowed them to 224 00:18:12,364 --> 00:18:23,582 to grow so quickly, but now that firms are comfortable with the cloud and all the big AMLaw firms are either there or making their way there, that additional friction, that air 225 00:18:23,582 --> 00:18:30,067 gap, now makes it challenging when firms want to build integrated experiences. 226 00:18:30,067 --> 00:18:40,589 When we show a client relationship page, we deploy in the client's tenant, so we can tap into their practice management solution, their DMS. 227 00:18:40,589 --> 00:18:45,860 experience management, again, rich attorney profiles for everybody that's spending time to those matters. 228 00:18:45,860 --> 00:18:55,664 We can pull into their analytics and show law firms never share whip, legal spend over time, all of these sorts of things. 229 00:18:55,664 --> 00:19:00,766 the current model, the current incumbent in that space, the architecture makes that challenging. 230 00:19:00,766 --> 00:19:07,097 we see a lot of opportunity there to do things differently, but 231 00:19:07,099 --> 00:19:11,287 What is your take on why it's been just file sharing? 232 00:19:11,287 --> 00:19:13,982 it those dynamics or are there other ones? 233 00:19:13,982 --> 00:19:15,784 I think that's a good point. 234 00:19:15,784 --> 00:19:20,928 think that probably the biggest reason is that the attorneys themselves aren't in those systems, right? 235 00:19:20,928 --> 00:19:27,891 So, they're certainly going to be reluctant to grant access to a platform that they're not themselves actually using on a day-to-day basis. 236 00:19:27,992 --> 00:19:37,309 Because the protection, can't really be overstated how sacrosanct like the attorney-client relationship is to partners. 237 00:19:37,309 --> 00:19:41,580 how protective partners can be over those relationships. 238 00:19:41,580 --> 00:19:48,620 Because obviously they're the lifeblood of the firms, but they're also the lifeblood of the individual partners who were working on them. 239 00:19:48,700 --> 00:19:55,671 so ensuring that you don't, that you control the narrative is very important to clients. 240 00:19:55,671 --> 00:19:59,211 Like you said, work in progress, like showing them stuff like was incomplete. 241 00:19:59,211 --> 00:20:03,131 You don't want them to kind of see your internal impressions, right? 242 00:20:03,131 --> 00:20:04,944 Like you might have an associate who, 243 00:20:04,944 --> 00:20:09,705 was working on a draft and you wouldn't want to show like, and they asked, they're like, is this really what we want? 244 00:20:09,705 --> 00:20:10,966 Or what about this other counter argument? 245 00:20:10,966 --> 00:20:15,527 And that might not be the narrative that you ultimately want to present to the client. 246 00:20:15,527 --> 00:20:24,246 And so it can cause uncertainty in your analysis if the, know, if you inadvertently share, you know, certain materials to, to a client. 247 00:20:24,246 --> 00:20:28,413 So I think there's been, there's a lot of protection around that relationship. 248 00:20:28,413 --> 00:20:33,028 And so when you add to the fact that like the, the systems you're mentioning of integrating, 249 00:20:33,028 --> 00:20:36,374 and being able to display, the partners themselves aren't really in those platforms. 250 00:20:36,374 --> 00:20:43,320 I think what's changed, least for Legora, is that we actually have partners and associates like you're in Legora daily, right? 251 00:20:43,320 --> 00:20:53,130 And so we're not coming from the outside and going to a partner and saying, hey, we want to share your content, your work product with your clients, and you're not using it. 252 00:20:53,130 --> 00:20:57,894 No, it's the partners and the associates who are using Legora who are saying, hey, 253 00:20:57,894 --> 00:21:03,227 we have this work product, is there a way that we can control how we share this with our clients? 254 00:21:03,227 --> 00:21:12,322 And so, and I would distinguish the kind of two use cases of, we have a final work product and we want to share that with the client and introduce them to the workspace versus like 255 00:21:12,322 --> 00:21:21,089 the more self-service productized legal services themselves, as they're rethinking the service delivery, making that obviously if it's self-service, you need some place where 256 00:21:21,089 --> 00:21:25,569 the clients are going to come in to perform the legal services or to have it be performed. 257 00:21:25,569 --> 00:21:31,593 ah But in both instances, it's the attorneys, partners, the associates are already in the platform using it. 258 00:21:31,593 --> 00:21:33,805 And then they want to introduce clients to that environment. 259 00:21:33,805 --> 00:21:35,477 So they're already comfortable with the environment. 260 00:21:35,477 --> 00:21:44,003 They just want to be able to know exactly, and this is from a product design perspective, what's really important, you know, to make sure that as we're building this product, that 261 00:21:44,003 --> 00:21:49,717 it is absolutely crystal clear what's being shared with the client at all times so that the partners feel comfortable. 262 00:21:49,717 --> 00:21:50,928 They're already comfortable in the environment. 263 00:21:50,928 --> 00:21:54,538 Now they're comfortable with, okay, they know exactly what they're sharing with whom. 264 00:21:54,538 --> 00:21:55,688 and what the clients can see. 265 00:21:55,688 --> 00:22:03,122 And so, you we allow them, for example, if you're sharing a workflow with clients, you can have them just see the output, but not see the reasoning. 266 00:22:03,122 --> 00:22:06,283 You can have them see the reasoning and the output, but not the prompts. 267 00:22:06,283 --> 00:22:06,623 Right? 268 00:22:06,623 --> 00:22:15,907 So we allow that granular control, which is particularly important as they productize their legal services, because they're basically taking their hundreds of years, know, or 269 00:22:15,907 --> 00:22:19,098 decades of knowledge and experience they have representing. 270 00:22:19,098 --> 00:22:23,330 mean, there's a reason why top tier firms command a premium in the market. 271 00:22:23,330 --> 00:22:27,954 um And one of those is that the cumulative experience they have handling a certain type of work. 272 00:22:27,954 --> 00:22:32,958 And so they're taking that and they're embedding that into technology, into these workflows. 273 00:22:32,958 --> 00:22:35,970 And so they don't, that becomes their IP, right? 274 00:22:35,970 --> 00:22:44,006 They're something that was in somebody's head and they're embedding it into a workflow and it is now firm IP that the firm wants to protect and needs to protect. 275 00:22:44,006 --> 00:22:50,018 And so to be able to ensure that as they're sharing the workflows, the work can get done, client can come in and say, 276 00:22:50,018 --> 00:22:59,228 run a contract playbook through the Legora portal, the client can have the playbook be run and they can get the output from uh Legora, but they don't necessarily see the playbook 277 00:22:59,228 --> 00:23:01,590 itself that is the IP of the firm. 278 00:23:01,590 --> 00:23:03,041 Yeah, that makes sense. 279 00:23:03,041 --> 00:23:05,813 Like how did, how should customers think about. 280 00:23:05,813 --> 00:23:16,229 Legora clients or other vendors clients when they make systems client facing and they decide to pivot at some point, right? 281 00:23:16,229 --> 00:23:21,482 Like they, they want, is there operational risk that they need to think about? 282 00:23:21,482 --> 00:23:27,946 Like, what if you, like, let's say you're a Legora client today and you share. 283 00:23:28,038 --> 00:23:38,593 through Lagora portal, a client really likes it and let's say the partner wants to switch to another platform or like how do you get that kind of pushback? 284 00:23:38,593 --> 00:23:42,665 Like, hey, if I show this to my clients and I'm kind of, we're getting hitched here. 285 00:23:42,665 --> 00:23:45,568 Do you ever get, have those sorts of conversations? 286 00:23:45,568 --> 00:23:48,270 Yeah, I mean, I guess it's an interesting question. 287 00:23:48,270 --> 00:23:53,882 So I think it highlights the importance of vendor selection, And like vendor durability. 288 00:23:53,882 --> 00:23:58,460 think one question that does come up in the legal tech space right now, there's a lot of players. 289 00:23:58,460 --> 00:24:04,565 uh It's every year at Iltta or Legal Week, there's tons of new players in the market. 290 00:24:04,565 --> 00:24:12,588 And with an industry that's happening so fast, everything's happening so fast and there's so many new market entrants in terms of new legal tech companies popping up all the time. 291 00:24:12,588 --> 00:24:16,411 And we're now starting to see some of them disappear. 292 00:24:16,411 --> 00:24:26,057 it causes a concern, obviously, on the part of firms, as well as their clients, about vendor durability and who they want to work with. 293 00:24:26,057 --> 00:24:34,893 particularly, where we've been thus far, think, been quite flexible, I think, in terms of like, if you're talking about just like a basic AI legal assistant, there's not a lot of 294 00:24:34,893 --> 00:24:36,834 risk or switching costs there. 295 00:24:37,052 --> 00:24:47,517 But as we really look to the future, know, as firms look to embed their knowledge experience into workflows, they have to be, it's incredibly important that they go with 296 00:24:47,517 --> 00:24:49,377 vendors that they trust. 297 00:24:49,377 --> 00:24:55,601 mean, we're basically asking customers to help us build the future of legal practice together. 298 00:24:55,601 --> 00:25:00,904 And as we do that, you know, it's very important who they select to build that future. 299 00:25:00,948 --> 00:25:06,651 And obviously there are a couple of things that clients or customers look to when deciding among vendors. 300 00:25:06,651 --> 00:25:08,532 I think financing is one of them. 301 00:25:08,532 --> 00:25:13,636 You want somebody who has sufficient uh capital to stand the test of time. 302 00:25:13,636 --> 00:25:16,919 Lawyers are notoriously peer obsessed. 303 00:25:16,919 --> 00:25:24,324 So looking to their peers and seeing what other firms are going with, gives them lot of confidence in terms of moving forward. 304 00:25:24,324 --> 00:25:26,513 But I think that the answer really is to just 305 00:25:26,513 --> 00:25:28,776 We're going to build the future together with these firms. 306 00:25:28,776 --> 00:25:32,139 And ultimately, if they're successful, we're going to be successful. 307 00:25:32,139 --> 00:25:38,165 And the hope would be is that they're successful, there's not going to be a need for them to move uh to another platform. 308 00:25:38,165 --> 00:25:43,389 What they're hoping to achieve out of this, in a lot of instances, is client stickiness. 309 00:25:43,389 --> 00:25:51,879 So they're hoping to entrench their client relationships via these tools and via these technologies. 310 00:25:51,879 --> 00:26:02,919 If they are the go-to platform for, like if you create like a self-service labor and employment portal for a client, and the client in-house counsel is using it daily to 311 00:26:02,919 --> 00:26:08,990 assess HR, to do compliance, to do customer investigations, to do reductions in forces, right? 312 00:26:08,990 --> 00:26:16,910 Like all these different aspects of labor and employment practice, if that can all be channeled through something like the Legora portal, you're going to create incredible 313 00:26:16,910 --> 00:26:19,646 client stickiness with that platform. 314 00:26:19,646 --> 00:26:24,606 And you wouldn't want to move at that point. 315 00:26:24,624 --> 00:26:24,954 Yeah. 316 00:26:24,954 --> 00:26:35,488 And you you had experience when you were on the law firm side with kind of like, it's interesting how many, so we don't have a ton of competition, thankfully, you know, on the 317 00:26:35,488 --> 00:26:41,231 intranet side, we had one competitor that's really kind of gone from the marketplace for the most part. 318 00:26:41,231 --> 00:26:49,296 And we're the only, it's either us or custom dev on the, on the extranet side, there's an incumbent in that space, which I talked about earlier. 319 00:26:49,296 --> 00:26:52,877 and you know, it's really, I see customers. 320 00:26:52,877 --> 00:26:55,919 debating, we buy it or build it? 321 00:26:56,039 --> 00:27:08,859 And it's such an easy, like we spent 14 years as Acrowire building with law firms and the ability for us to take what we built and hand over to them where they could manage the 322 00:27:08,859 --> 00:27:18,415 care and feeding and the enhancements and the support and the training and keeping their ear out, ear to the ground for any integrations that would implement breaking changes. 323 00:27:18,415 --> 00:27:21,237 Like they're just, it's such a high bar. 324 00:27:21,385 --> 00:27:28,337 And a lot of times law firms will go, but it's only going to cost me, you know, 600,000 to build this thing out. 325 00:27:28,337 --> 00:27:32,708 And I'm like, yeah, that's, that's the part that has a discrete beginning and end. 326 00:27:32,708 --> 00:27:37,070 It's the ongoing costs that really drive TCO. 327 00:27:37,070 --> 00:27:41,282 sometimes four to five X your initial development costs. 328 00:27:41,282 --> 00:27:50,494 So like, it blows my mind that I still have to have those conversations, but the reality is I do, because some, some firms have 329 00:27:50,540 --> 00:27:52,460 dev teams and guess what? 330 00:27:52,460 --> 00:27:54,380 Devs love to dev, right? 331 00:27:54,380 --> 00:27:54,860 I did. 332 00:27:54,860 --> 00:27:56,380 I used to be a developer. 333 00:27:56,520 --> 00:27:59,780 Um, and you know, it's, that's what you want to do. 334 00:27:59,780 --> 00:28:09,880 You want to flex those muscles, but like, I don't know, do you run into, how, how do you, know, in your experience, not just today with the Gora, but like, how have you thought 335 00:28:09,880 --> 00:28:13,351 about that problem kind of build versus buy in the law firm context? 336 00:28:13,351 --> 00:28:22,052 Yeah, so, you know, when I started, you know, really 2013, like started working with firm IT, you know, the legal tech market was not as mature as it is today. 337 00:28:22,052 --> 00:28:25,312 So I mean, guys, InfoDash was not around at that time. 338 00:28:25,312 --> 00:28:34,472 You know, so the, if I were to start down this journey today, or if there were an associate at a firm who were to start down the journey today where I was, like, there's no 339 00:28:34,472 --> 00:28:37,832 world in which they would have built the software internally. 340 00:28:37,984 --> 00:28:40,224 Not to say you wouldn't do some internal development. 341 00:28:40,224 --> 00:28:44,504 I think there's absolutely a role for certain types of development within firms, even today. 342 00:28:44,504 --> 00:28:48,904 But you certainly wouldn't have built out the matter management platform as I did. 343 00:28:48,904 --> 00:28:58,204 What we would have done instead is we would have got one of the top tier matter management type platforms that exists on the market that are fully customizable. 344 00:28:58,204 --> 00:29:01,784 We would have bought that or licensed that and then we would have just customized. 345 00:29:02,124 --> 00:29:06,095 If I were to advise somebody today, that's the route you go down. 346 00:29:06,095 --> 00:29:11,495 build a fully customizable, you don't internally build a customizable amount of management platform yourself. 347 00:29:11,955 --> 00:29:16,655 Because you could, mean, you could, but it's it's not, there's just too many options now. 348 00:29:16,655 --> 00:29:20,435 So the calculus has changed from a decade ago. 349 00:29:20,435 --> 00:29:23,115 Now there's lots of great solutions to the market. 350 00:29:23,135 --> 00:29:27,655 Certainly, as I said, I did a client, I did the extra platform at Morgan Lewis. 351 00:29:27,655 --> 00:29:30,595 And today it wouldn't make sense to do that. 352 00:29:30,595 --> 00:29:32,835 InfoDash wasn't in the extra net space at that time, right? 353 00:29:32,835 --> 00:29:35,568 So it wouldn't make sense to do that today. 354 00:29:35,568 --> 00:29:39,212 I do, you know, it's interesting to say devs like to dev, that's absolutely true. 355 00:29:39,212 --> 00:29:42,356 And the flip side of that is they don't really like to do maintenance and support. 356 00:29:42,356 --> 00:29:50,103 And so, that's, they want to work on the new thing and they don't want to, you know, uh tend to the legacy stuff. 357 00:29:50,103 --> 00:29:54,366 So that creates a, you know, certainly like it increases the total cost of ownership of this, these tools. 358 00:29:54,366 --> 00:29:58,030 And of course, if you built something fully custom, you're almost wedded to the people who built it, right? 359 00:29:58,030 --> 00:30:00,350 Because those people are the only ones who know. 360 00:30:00,350 --> 00:30:02,221 how that system operates. 361 00:30:02,221 --> 00:30:06,764 And it's difficult to, uh you can't just go off the street and find somebody who knows how to do it. 362 00:30:06,764 --> 00:30:13,607 Whereas like, you if you were to use, just take a non-legal example, if you were to build like a custom solution on top of Salesforce, you know, there's a whole marketplace of 363 00:30:13,607 --> 00:30:19,433 people out there who have Salesforce dev skills, who could support your application if, you know, somebody were to leave your company. 364 00:30:19,433 --> 00:30:23,916 So there's certainly lots of downstream implications of doing custom dev. 365 00:30:23,916 --> 00:30:26,057 The stuff that does make sense, 366 00:30:26,064 --> 00:30:36,437 you know, in terms of custom development in firms, integration work makes a lot of sense given that uh each firm's systems, know, even if they use like say one of the leading, you 367 00:30:36,437 --> 00:30:44,551 know, timing and billing systems or, you know, the, each one is so customized to each firm, you know, the amount of complexity there is something that most people, unless 368 00:30:44,551 --> 00:30:52,439 you've been kind of on the database side of things, you don't know how complicated like timing and billing systems are really, you know, like the matter of practice management 369 00:30:52,439 --> 00:30:54,095 systems inside of firms. 370 00:30:54,095 --> 00:30:56,578 each one is highly bespoke to that firm. 371 00:30:56,578 --> 00:31:06,816 And so the work that you need to do to say integrate a new business intake platform with another like matter management system is highly specific, is not something you're going to 372 00:31:06,816 --> 00:31:07,938 be able to find off the shelf. 373 00:31:07,938 --> 00:31:11,030 So that kind of work really makes sense. 374 00:31:11,030 --> 00:31:20,436 We do have the conversation kind of today in my role where I'm talking with firms all the time and I'll show them something you can do with Legar and they'll say, 375 00:31:20,496 --> 00:31:21,996 we built that. 376 00:31:21,996 --> 00:31:24,296 then of course I asked them, like, well, how long did it take you to build that? 377 00:31:24,296 --> 00:31:27,216 And they'll say, oh, well, it took us like four months and we did it like a year ago. 378 00:31:27,216 --> 00:31:28,596 And it's like, awesome. 379 00:31:28,596 --> 00:31:34,736 know, that, you probably that was something that Lagora didn't do a year ago, but you know, today I did it in Lagora in an hour. 380 00:31:34,896 --> 00:31:42,816 And so if once things become commercially available, right, like once you can use commercial software and you can just take off the shelf commercial software and customize 381 00:31:42,816 --> 00:31:48,936 it to meet your use case, it no longer makes sense to dedicate development resources internally to reinvent the wheel. 382 00:31:49,460 --> 00:31:59,610 Instead, it makes sense to dedicate those resources to things that are highly practice or matter specific that uh you wouldn't be able to find commercially available. 383 00:31:59,610 --> 00:32:07,167 things that, uh you know, and then it looks, makes sense to integrate those with a platform like Legora via something like MCP, right? 384 00:32:07,167 --> 00:32:14,696 So that way we can handle sort of commercially available use cases, document review, document traction, things like word add-in. 385 00:32:14,696 --> 00:32:22,292 we can build things like great user experiences, interfaces, permissioning and security and all those tasks that aren't necessarily the most exciting. 386 00:32:22,292 --> 00:32:27,155 And then firms can focus on what sort of is their competitive advantage, which is their knowledge and experience, right? 387 00:32:27,155 --> 00:32:36,389 And so they can focus on building technology that really lends to their own strengths and let us at, in the tech vendor space, let us build great software. 388 00:32:36,435 --> 00:32:37,206 Yeah. 389 00:32:37,206 --> 00:32:38,506 Yeah, that makes sense. 390 00:32:38,506 --> 00:32:45,592 How do you think firms, what is the differentiation equation going to look like, you know, five years from now? 391 00:32:45,592 --> 00:32:45,932 Right? 392 00:32:45,932 --> 00:32:52,436 Like you can't buy an off the shelf solution and differentiate yourself based on that tool. 393 00:32:52,436 --> 00:33:02,925 You really kind of have to, you know, I have a theory about this, that law firms are going to, their special sauce is all of the knowledge and documents that created winning 394 00:33:02,925 --> 00:33:04,776 outcomes for their clients. 395 00:33:04,872 --> 00:33:05,132 Right. 396 00:33:05,132 --> 00:33:08,104 And how do they use that in a wholesale way today? 397 00:33:08,104 --> 00:33:11,305 Like, Legora is a, is a hosted solution, right? 398 00:33:11,305 --> 00:33:17,589 You're kind of bringing your data to AI versus bringing AI to your data. 399 00:33:17,589 --> 00:33:22,162 Like how, how are firms, how are firms going to differentiate themselves? 400 00:33:22,162 --> 00:33:25,194 Cause again, off the shelf tools by themselves aren't going to do it. 401 00:33:25,194 --> 00:33:32,752 Maybe it's through the workflows that you're talking about, but yeah, how is firm A going to differentiate from firm B? 402 00:33:32,752 --> 00:33:34,032 three to five years from now. 403 00:33:34,032 --> 00:33:38,184 Yeah, so I think what you're getting at though is that there's kind of a short and a long term. 404 00:33:38,184 --> 00:33:46,407 So in the short term, there is some early mover advantages of people who, you know, who lean into the tools, you know, not just licensing, but like who really get up to speed and 405 00:33:46,407 --> 00:33:48,988 integrate tools throughout their firms today, right? 406 00:33:48,988 --> 00:33:55,431 Because they have a strategic advantage over their competitors, say a trial, because they're using AI and their competitors aren't using AI. 407 00:33:55,431 --> 00:34:02,084 But certainly, uh once this kind of short term, once, you know, once everybody has a legora, for example, right? 408 00:34:02,150 --> 00:34:05,474 and it's become the incumbent solution or has some other tools. 409 00:34:05,474 --> 00:34:10,639 It won't be, you're right to say that you can't, there won't be a competitive advantage because everybody has it, right? 410 00:34:10,639 --> 00:34:18,467 I mean, so this will be sort of like, if we go back and kind of by analogy, you know, there were some competitive advantages early on with uh say legal research, electronic 411 00:34:18,467 --> 00:34:19,708 legal research, right? 412 00:34:19,708 --> 00:34:26,578 Early on, if you were a firm who leaned into electronic legal research, you were able to do more comprehensive legal research than your competitors were. 413 00:34:26,578 --> 00:34:35,058 that then gave you some strategic advantages at trial because you were able to find that one case that was on point or that case that rebuts what you're, know, that, that 414 00:34:35,058 --> 00:34:38,838 distinguished the precedent on which your opponent was relying. 415 00:34:38,838 --> 00:34:42,138 And so you're able to use, you gain a strategic advantage through that. 416 00:34:42,138 --> 00:34:46,498 But that then became very quickly because this is like a very competitive market. 417 00:34:46,578 --> 00:34:51,338 You know, once you're using it, once kind of the cat's out of the bag on electronic legal research, everybody's doing it. 418 00:34:51,338 --> 00:34:52,418 It just becomes the new normal. 419 00:34:52,418 --> 00:34:54,058 Like everybody has to do it. 420 00:34:54,074 --> 00:34:57,867 So there's no longer really any competitive advantage to using electronic research. 421 00:34:57,867 --> 00:34:59,199 It's just the baseline, right? 422 00:34:59,199 --> 00:35:06,185 And so in some ways, that will be the case with uh AI as well, is that like it will just become the new baseline. 423 00:35:06,185 --> 00:35:13,791 In the meantime, though, those firms that lean into it will, you know, do have a chance to get ahead and to take advantage of that, kind of seize the moment and get ahead. 424 00:35:13,791 --> 00:35:16,763 So I think that, you know, what will be their competitive advantage? 425 00:35:16,763 --> 00:35:19,795 mean, the good news in some ways is that it won't be the software, right? 426 00:35:19,795 --> 00:35:27,655 If they don't have to worry about like once everybody has AI, once they've, it's been the law schools are teaching it and all the associates know and the partners are using it on a 427 00:35:27,655 --> 00:35:28,915 day to day basis. 428 00:35:28,935 --> 00:35:32,095 It won't be the tech that you buy that's going to be your competitive advantage. 429 00:35:32,095 --> 00:35:36,034 It's going to go back to being the core things that you went to law school for, right? 430 00:35:36,034 --> 00:35:44,235 Look at the core things that firms use to differentiate today, which is like their people, their experience, their knowledge, their work product, their relationships, their brand, 431 00:35:44,235 --> 00:35:47,095 like those core things they use to differentiate themselves today. 432 00:35:47,095 --> 00:35:49,236 Those will be the enduring things. 433 00:35:49,236 --> 00:35:55,125 that remain once AI has kind of reshaped the landscape and however it's going to do it. 434 00:35:55,125 --> 00:35:57,607 Yeah. 435 00:35:57,607 --> 00:36:08,255 what about the amount of services that inside teams are going to work that once went to a law firm that they're going to manage internally? 436 00:36:08,255 --> 00:36:12,117 That is going to displace a certain amount of law firm revenue. 437 00:36:12,117 --> 00:36:14,349 And that's not a Lagora problem or a Harvey problem. 438 00:36:14,349 --> 00:36:15,850 That's the tech. 439 00:36:15,850 --> 00:36:17,241 The tech is going to do this. 440 00:36:17,241 --> 00:36:20,978 And law firms just need to come to grips with, I think that they're 441 00:36:20,978 --> 00:36:25,370 credible estimates that 30 % of associate hours today are gone. 442 00:36:25,370 --> 00:36:33,903 Based on where the tech is today, you can take 30 % of associate hours and essentially tech is going to be delivering on that. 443 00:36:33,903 --> 00:36:38,534 It doesn't matter if it's your platform or the platform down the street. 444 00:36:38,534 --> 00:36:44,197 Firms have to move up the value chain, which creates a lot of other challenges. 445 00:36:44,197 --> 00:36:47,038 Well, that's how you train lawyers. 446 00:36:47,038 --> 00:36:48,679 Is the grunt work? 447 00:36:48,679 --> 00:37:00,185 You throw them in a room with a box of documents and you review them and you have a partner parachute in and and Provide guidance and this this tech kind of blows that all up 448 00:37:00,185 --> 00:37:10,632 So, you know, it creates it creates challenges like okay the leverage model associate development You know potentially having a revenue impact 449 00:37:11,043 --> 00:37:20,358 Do you think that there's sufficient, because when I talk to some people, they point to Javan's paradox and go, there's a ton of unmet need and that's going to more than offset. 450 00:37:20,358 --> 00:37:25,041 And that could be true to a certain extent, but demand isn't infinite. 451 00:37:25,041 --> 00:37:33,905 So it's unclear at this point how far up the complexity spectrum AI is going to make its way and how fast. 452 00:37:33,905 --> 00:37:35,907 We don't know the answers to that. 453 00:37:35,907 --> 00:37:43,929 In fact, when you talk to credible sources at Frontier Labs today, they've kind of reached their limits on scaling, right? 454 00:37:43,929 --> 00:37:55,245 Like we've consumed all the publicly available data that's out there and just throwing more parameters at a model is starting to have diminishing returns. 455 00:37:55,245 --> 00:38:04,670 So there's a lot of talk about like us entering into a new age of research where, you know, scaling sucked the air out of the room for like the last four years. 456 00:38:04,826 --> 00:38:14,075 And there really wasn't an emphasis in resources placed on research and new approaches around engineering. 457 00:38:14,075 --> 00:38:17,619 Now with inference, we have a path to make things better. 458 00:38:17,619 --> 00:38:19,551 even there, feel like we're... 459 00:38:19,551 --> 00:38:21,809 Do you think we're gonna need a new... 460 00:38:21,809 --> 00:38:25,813 Are LLMs gonna be the answer in three to five years? 461 00:38:25,813 --> 00:38:27,044 Do you see the trajectory? 462 00:38:27,044 --> 00:38:29,115 And I know it's hard to say for sure, but... 463 00:38:29,173 --> 00:38:38,317 Just based on what you know, do you see LLMs as being the future state in three to five years or might it be something different? 464 00:38:38,769 --> 00:38:45,513 Yeah, I mean, it's probably beyond my, know, I'm not a researcher in the LLM space, so I don't know the AI space more broadly. 465 00:38:45,513 --> 00:38:50,176 So I am probably outside of my, you know, was kind of above my pay grade, so to speak. 466 00:38:50,176 --> 00:38:54,464 I mean, I would imagine it's hard for me to based on what I know in my experience of it. 467 00:38:54,464 --> 00:38:59,162 You know, there are people talk about world models being kind of the next frontier of AI. 468 00:38:59,162 --> 00:39:00,152 And I can see that. 469 00:39:00,152 --> 00:39:05,275 But I still kind of haven't gotten a very clean answer as why a why LLMs couldn't 470 00:39:05,275 --> 00:39:10,690 be the source data for creating synthetic data for that, to create scenarios that world models could then be trained on. 471 00:39:10,690 --> 00:39:16,473 So I would find it very perplexing if what we have today isn't a stepping stone to whatever's next. 472 00:39:16,473 --> 00:39:22,237 um Is a new breakthrough, a new paradigm going to be necessary to get to that new location? 473 00:39:22,237 --> 00:39:23,908 Yeah, to that new tier, likely. 474 00:39:23,908 --> 00:39:30,482 You're kind of referring to basically foundational models plateauing, the idea that scaling isn't going to get us more 475 00:39:30,482 --> 00:39:34,302 I think, yeah, it's quite possible. 476 00:39:34,382 --> 00:39:43,262 When I think about for legal though, I think that even if foundational models plateaued, we're barely scratching the surface, just in terms of just the number of attorneys who are 477 00:39:43,262 --> 00:39:43,722 using it. 478 00:39:43,722 --> 00:39:50,622 I I didn't see a stat to like half of all attorneys in firms of 500 or more are using AI last year. 479 00:39:50,622 --> 00:39:53,093 Okay, well, like they used it, what, once? 480 00:39:53,093 --> 00:39:55,769 The amount of adoption is still... 481 00:39:55,769 --> 00:39:57,180 uh relatively low. 482 00:39:57,180 --> 00:39:59,561 mean, obviously it's highly unequal. 483 00:39:59,561 --> 00:40:03,772 There are some people who using it every single day in some firms who have really high usage. 484 00:40:03,772 --> 00:40:13,199 But just industry-wide, uh we're still early days in terms of just total number of people who are using it and the value they're getting out of it. 485 00:40:13,199 --> 00:40:22,244 Okay, somebody might be using it to proofread an email, but that's really low-hanging fruit in terms of what these AI models today are capable of doing. 486 00:40:22,244 --> 00:40:24,914 So from a legal AI perspective, 487 00:40:24,914 --> 00:40:29,898 whether the models plateau or not, there's still plenty of value that we're going to be able to get. 488 00:40:29,898 --> 00:40:31,979 I don't think the models are going to plateau anytime soon. 489 00:40:31,979 --> 00:40:36,341 I do think there will be those breakthroughs. 490 00:40:36,341 --> 00:40:45,387 I don't have any specific insight into what those will be, but I think that given how much progress has been made, it'd be kind of silly to bet against them, to bet against them 491 00:40:45,387 --> 00:40:46,568 being a breakthrough. 492 00:40:46,568 --> 00:40:50,906 And we're making a civilizational bet here on AI, right? 493 00:40:50,906 --> 00:40:56,546 the amount of money that's being invested in AI is probably the largest investment we've ever made as a species. 494 00:40:56,646 --> 00:41:03,046 And so, you know, this is, and the payoff is potentially going to be the most important technology since, you know, the invention of writing. 495 00:41:03,366 --> 00:41:07,125 So it's, you know, really exciting. 496 00:41:07,125 --> 00:41:13,146 And I think that even if we don't hit AGI or whatnot, I think we're going to still see incredible things come out of this. 497 00:41:13,506 --> 00:41:20,822 It's just, you know, in the short term, like, you know, we, it seems slow, like from GBT four to five and then five to 5.1 and two, like, 498 00:41:20,822 --> 00:41:24,062 It seems small, but if you just zoom out, it's accelerating. 499 00:41:24,062 --> 00:41:26,716 It really is impressive, the rate of... 500 00:41:26,716 --> 00:41:29,708 We just get used to things so quickly, right? 501 00:41:29,708 --> 00:41:31,029 And so we're not really impressed. 502 00:41:31,029 --> 00:41:34,080 But you raised a number of other issues too. 503 00:41:34,080 --> 00:41:36,902 So I think Jevin's paradox, I think that's probably right. 504 00:41:36,902 --> 00:41:39,162 I agree with you, is certainly not infinite. 505 00:41:39,162 --> 00:41:47,019 I was reading an article the other day about radiology and there were predictions really even before GNI, like in terms of predictive coding of... 506 00:41:47,019 --> 00:41:47,763 uh 507 00:41:47,763 --> 00:41:51,483 or of traditional machine learning on x-rays. 508 00:41:51,483 --> 00:41:56,557 People were predicting this back in 2016 that there would be like, don't go to, don't study radiology. 509 00:41:56,557 --> 00:42:01,449 Like that field's gonna die because the computers are going to, uh you just plant all that. 510 00:42:01,449 --> 00:42:11,293 There's more demand today for radiology than there ever has before because simply as the cost goes down to examine an x-ray, the number of doctors who are then sending things off 511 00:42:11,293 --> 00:42:13,764 to get analyzed is up, right? 512 00:42:13,764 --> 00:42:15,247 So like there are now... 513 00:42:15,247 --> 00:42:18,367 greater demand for radiologists than there's ever been before. 514 00:42:18,367 --> 00:42:22,127 And radiological services are being demanded higher than they've ever been before. 515 00:42:22,287 --> 00:42:24,127 And will that continue forever? 516 00:42:24,127 --> 00:42:25,807 Well, probably not, because nothing will. 517 00:42:25,807 --> 00:42:28,927 But I think that it certainly will continue for a long time. 518 00:42:28,927 --> 00:42:38,127 So even in another domain, just as to benchmark what we're thinking in terms of legal, that's a very highly specialized task that people thought was going to be automated very 519 00:42:38,127 --> 00:42:38,567 quickly. 520 00:42:38,567 --> 00:42:43,681 And what it turned out there, which I think is very relevant to legal by analogy, is that 521 00:42:43,681 --> 00:42:49,053 The mistakes that humans make, it just so happens that they're the kind of things that AI is really good at catching. 522 00:42:49,053 --> 00:42:52,954 And the mistakes that AI makes, it turns out that humans are actually really good at catching. 523 00:42:52,954 --> 00:42:58,777 And so when you combine both humans plus AI, it's going to be better than either one on its own. 524 00:42:58,777 --> 00:43:04,419 There is so much demand for legal services that never gets met because attorneys are costly. 525 00:43:04,419 --> 00:43:08,716 Obviously, there are different segments of the legal market. 526 00:43:08,716 --> 00:43:17,816 But there are plenty of businesses where you never bother to call an attorney because even in Fortune 500 companies, because you're like, well, it's not going to be worth the point 527 00:43:17,816 --> 00:43:21,696 two or the point three that they're going to take or the couple of hours they're going to bill me. 528 00:43:21,796 --> 00:43:27,736 So people are just running the risk all the time by not calling lawyers and not consulting with lawyers. 529 00:43:27,736 --> 00:43:37,452 And so once you can lower the cost of getting advice on a given question, the number of times people are going to call their attorneys is going to go up. 530 00:43:37,452 --> 00:43:40,632 And that's just on the compliance advisory space. 531 00:43:40,732 --> 00:43:48,612 Once we go to the litigation space, it's kind of the game theory there is actually even more interesting because it's like, what does the world look like if everybody you were to 532 00:43:48,612 --> 00:43:53,132 give like an AI attorney, or if you were able to power all the new law school grads? 533 00:43:53,132 --> 00:43:57,332 Because one thing that's interesting is law school attendance is up at an all time high this year. 534 00:43:57,332 --> 00:44:00,432 The number of people who are looking to go to law school is at an all time high. 535 00:44:00,432 --> 00:44:04,292 So let's assume we have some of the largest law school graduating classes ever. 536 00:44:04,292 --> 00:44:05,812 What are those people going to do? 537 00:44:05,872 --> 00:44:13,199 Well, maybe with AI, they're going be able to go help meet the long tail of unmet needs of all the people out there who currently can't afford attorneys. 538 00:44:13,199 --> 00:44:21,037 Well, so now what does the world look like when all of a sudden everybody, or at least you've doubled the amount of people who are able to afford legal services? 539 00:44:21,037 --> 00:44:23,239 Well, now they're able to sue. 540 00:44:23,239 --> 00:44:27,002 When they wouldn't have been able to sue before, they're able to vindicate their rights in ways they weren't able to before. 541 00:44:27,002 --> 00:44:31,098 And that's then going to cause a massive increase in demand for defense attorneys. 542 00:44:31,098 --> 00:44:36,178 to then defend all the companies who are now being sued by all these new AI enabled attorneys. 543 00:44:36,318 --> 00:44:44,409 So a lot of these things, the game theory of it is that as demand on one side of the V increases, the demand will go up on the other side. 544 00:44:44,409 --> 00:44:51,309 So I don't think we're going to see, you say 30 % of associate hours can be replaced with AI. 545 00:44:51,709 --> 00:44:54,849 On a task level, that may be true. 546 00:44:54,909 --> 00:44:58,553 But it certainly doesn't necessarily follow that like they're 547 00:44:58,553 --> 00:45:04,813 going to you're going need a third of your associates or that you know that that time won't get repurposed. 548 00:45:04,813 --> 00:45:14,053 And you know, on the training thing, there's a myth, I think there certainly is some amount of work, you you know, it's like how I learned to be an attorney and how you know, 549 00:45:14,053 --> 00:45:17,353 most of the partners out there learn to be attorneys, like they did it through grinding. 550 00:45:17,353 --> 00:45:21,373 But like, if we're honest with ourselves, like a lot of that work wasn't productive, right? 551 00:45:21,373 --> 00:45:23,273 A lot of it was just a time sink. 552 00:45:23,273 --> 00:45:25,073 It was just a way to get a lot of hours. 553 00:45:25,073 --> 00:45:27,403 It was a way to generate leverage for a firm. 554 00:45:27,403 --> 00:45:31,016 And it wasn't necessarily something that really trained us. 555 00:45:31,016 --> 00:45:33,498 And in fact, oftentimes it inhibits you. 556 00:45:33,498 --> 00:45:38,903 you're so busy doing the grunt work that you don't even have time to like learn the higher level skills. 557 00:45:38,903 --> 00:45:41,145 And so it's actually slowing your progress. 558 00:45:41,145 --> 00:45:49,612 And so as AI can handle more of that underlying work, you're able to then focus on the higher level task, both, you know, even as you're learning. 559 00:45:49,612 --> 00:45:54,986 And I guess the other thing too is that, you know, how are we going to learn or train associates? 560 00:45:54,986 --> 00:45:56,117 Like if it's like, 561 00:45:56,229 --> 00:46:01,671 ah if it's helping on a lot of these lower-level tasks, mean, the AI is clearly going to be helping us learn too, right? 562 00:46:01,671 --> 00:46:06,212 AI done, deployed properly, explains its results, right? 563 00:46:06,212 --> 00:46:09,003 It explains where it's getting its analysis. 564 00:46:09,003 --> 00:46:13,075 so oftentimes when people ask me this, particularly in law firm context, they're like, well, how are we going to train associates? 565 00:46:13,075 --> 00:46:16,377 And it's like, well, it presumes you do a good job of training today. 566 00:46:16,377 --> 00:46:20,646 Training in law firms today is highly unequal. 567 00:46:20,646 --> 00:46:25,606 Not all associates get the same mentorship opportunities as others, right? 568 00:46:25,606 --> 00:46:31,506 It's like, you're really, you know, it's very unequal who gets access to the mentorship opportunities. 569 00:46:31,566 --> 00:46:43,006 And so if the AI can help us kind of bridge that gap and you level that playing field where everybody has access to AI in a firm and they can get, the AI can have infinite time 570 00:46:43,006 --> 00:46:49,126 and infinite patience to explain how to do a due diligence, how to analyze a contract, how to write a brief. 571 00:46:49,126 --> 00:46:57,746 why, you know, when you, when I was an associate, I would get red lines of my briefs back, but there's no explanations as to why I had made a mistake or why that paragraph needed to 572 00:46:57,746 --> 00:46:59,746 be cut down or moved around. 573 00:46:59,806 --> 00:47:08,266 But with something like LaGaurat today, like we explain every single edit and there's detailed explanations as to why LaGaurat recommends that edit be made. 574 00:47:08,266 --> 00:47:11,526 And that's far more time and attention than like any partner would ever have. 575 00:47:11,526 --> 00:47:18,714 So I really think the AI is going to be, you know, there might be some bumps along the road, but I think the AI is actually going to be used to 576 00:47:18,714 --> 00:47:21,697 help us train some of the best attorneys we've ever seen. 577 00:47:21,806 --> 00:47:22,376 Yeah. 578 00:47:22,376 --> 00:47:22,627 Yeah. 579 00:47:22,627 --> 00:47:31,113 I don't disagree that there are much more effective methods than the traditional, you know, legacy method that we've used to train attorneys. 580 00:47:31,113 --> 00:47:35,656 There are absolutely more efficient and effective ways to do it. 581 00:47:35,656 --> 00:47:38,309 So I think I completely agree with you there. 582 00:47:38,309 --> 00:47:41,092 well, this has been a fantastic conversation, Kyle. 583 00:47:41,092 --> 00:47:47,366 like I said, you and I been chewing the fat in the, on the, the legal collaboration space for quite some time. 584 00:47:47,366 --> 00:47:48,975 So it's, uh 585 00:47:48,975 --> 00:47:52,567 I really appreciate you spending a few minutes with us this afternoon. 586 00:47:52,731 --> 00:47:54,422 Yeah, like I said, was a lot of fun. 587 00:47:54,422 --> 00:47:54,943 Awesome. 588 00:47:54,943 --> 00:47:55,394 All right. 589 00:47:55,394 --> 00:47:57,177 Well, listen, have a good rest of your day. 590 00:47:57,177 --> 00:47:57,958 We'll chat again soon. 591 00:47:57,958 --> 00:47:59,571 All right. 592 00:47:59,571 --> 00:48:00,142 Sure. 00:00:01,772 Kyle Poe, how are you this afternoon? 2 00:00:01,772 --> 00:00:05,163 I'm doing awesome. 3 00:00:05,163 --> 00:00:08,225 This is actually the third podcast I've recorded today. 4 00:00:08,225 --> 00:00:10,916 um Yeah, today's a busy day. 5 00:00:10,916 --> 00:00:18,871 Normally I do like one a week, but we had a ton of travel in the late fall, so we burned through our backlog. 6 00:00:18,871 --> 00:00:24,244 Normally we keep a big backlog, so I'm rebuilding the pipeline again, so to speak. 7 00:00:24,244 --> 00:00:25,335 Well, it's great to have you on. 8 00:00:25,335 --> 00:00:32,040 You and I got to know each other before, back when you were in the, still on the law firm side, you've now moved to the vendor side. 9 00:00:32,040 --> 00:00:34,842 And it's great to finally have you on the show. 10 00:00:34,842 --> 00:00:35,962 We got a great agenda. 11 00:00:35,962 --> 00:00:42,606 Before we jump in, why don't you just do a quick introduction and tell us who you are, what you do and where you do it. 12 00:00:42,984 --> 00:00:43,754 Sure. 13 00:00:43,754 --> 00:00:46,886 So Kyle Poe, VP of legal innovation and strategy here at Legora. 14 00:00:46,886 --> 00:00:53,589 So former practicing attorney, I practiced law for 14 years, most of which at Morgan Lewis, where I was an associate and then a partner there. 15 00:00:53,589 --> 00:01:01,192 uh My practice was primarily mass tort and product liability litigation, but I also dabbled in labor and employment as well as just other commercial litigation. 16 00:01:01,212 --> 00:01:05,344 Really early in my career, I got involved in the tech side of legal practice. 17 00:01:05,344 --> 00:01:09,588 So building custom technology solutions internally at Morgan Lewis. 18 00:01:09,588 --> 00:01:15,068 which led me ultimately to using AI at the firm and using it for my own practice. 19 00:01:15,068 --> 00:01:22,048 I got to see firsthand how powerful AI was and that gave me the conviction that this was really going to be a game changer. 20 00:01:22,048 --> 00:01:31,348 I had been doing legal tech and building software in the firm for a decade and nothing before that was as powerful or as compelling as AI had been. 21 00:01:31,768 --> 00:01:38,588 I wanted to be part of this wave and so I decided to leave at that time and go on to the vendor side as you said. 22 00:01:38,866 --> 00:01:40,238 led me to here at Legora. 23 00:01:40,238 --> 00:01:48,384 And what I do here is I work with customers, help them understand the technology as well as adapt it for their own practices as well as their own business. 24 00:01:48,384 --> 00:01:55,929 So really thinking about the future of legal practice and how this is changing not only what they're doing on a day-to-day basis, but also kind of the next level in terms of how 25 00:01:55,929 --> 00:02:00,232 they're reshaping their pricing, their packaging, their delivery of legal services using AI. 26 00:02:00,648 --> 00:02:01,669 Yeah, you know, it's interesting. 27 00:02:01,669 --> 00:02:06,393 You don't see many partners that make the transition into the tech side of the business. 28 00:02:06,393 --> 00:02:07,565 Like what drove that? 29 00:02:07,565 --> 00:02:08,937 Was it just your passion? 30 00:02:08,937 --> 00:02:14,477 Yeah, I I had seen, I mean, there have been a couple of us who have made that jump. 31 00:02:14,617 --> 00:02:24,337 I think it's, it is unprecedented, but I think this is we sort of live in unprecedented times, you know, is AI, there's nothing like AI in terms of the breadth of experience, you 32 00:02:24,337 --> 00:02:28,437 know, I had been doing innovation internally, as I said at the firm. 33 00:02:28,477 --> 00:02:31,717 And, you know, I did it originally for my own master of practice. 34 00:02:31,717 --> 00:02:35,497 And then I got a reputation as being like the tech enabled lawyer. 35 00:02:35,497 --> 00:02:38,121 And so people would come to me, you know, I would speak to 36 00:02:38,121 --> 00:02:40,141 you know, partners throughout the firm. 37 00:02:40,141 --> 00:02:42,101 And they would ask like, hey, can you help? 38 00:02:42,101 --> 00:02:48,461 And I built up a team of data analysts and engineers and developers, a team of about 10 by the time I left. 39 00:02:48,461 --> 00:02:50,861 And so I had resources to deploy. 40 00:02:51,101 --> 00:02:55,661 But most of the time when I would talk to partners and I would ask them, okay, you know, what do you do on a day-to-day basis? 41 00:02:55,661 --> 00:02:57,641 You know, how's your business structured? 42 00:02:57,641 --> 00:02:58,541 What are your associates doing? 43 00:02:58,541 --> 00:02:59,901 Your mid-levels doing? 44 00:03:00,041 --> 00:03:01,461 What are your pain points? 45 00:03:01,681 --> 00:03:05,673 Most of the time, like 90 % of the time, I would come back to them and say, you know, 46 00:03:05,673 --> 00:03:18,033 whatever I'm building, which is a lot of databases, matter management, statistical analytics, most of those approaches, kind pre-gen AI technologies, weren't really adept at 47 00:03:18,033 --> 00:03:21,093 helping the vast majority of the partners that I would talk to. 48 00:03:21,093 --> 00:03:29,293 And so for the most of the time, I would just tell them, what I have and what I'm able to do isn't really going to make a huge impact for your practice. 49 00:03:29,293 --> 00:03:34,633 Basically, before gen AI, for you to have a good use case using pre-gen AI technologies, 50 00:03:34,633 --> 00:03:39,693 You need to have a practice that had a lot of repeatability to it, that had a lot of scale to it. 51 00:03:39,693 --> 00:03:48,313 Often you need to find a practice that already sort of had the right business model in place, say like a fixed fee or alternative fees or contingency fees, for it to really make 52 00:03:48,313 --> 00:03:50,863 sense to deploy pre-gen AI technologies. 53 00:03:50,863 --> 00:03:54,403 And so when GenAI came out and I started using it, it was just so clear. 54 00:03:54,403 --> 00:03:56,443 mean, look, computers learn to speak English. 55 00:03:56,443 --> 00:04:00,663 They learn to speak natural language for the first time with GenAI and LLMs. 56 00:04:00,803 --> 00:04:03,463 the basically 90, there's 90 % of 57 00:04:03,463 --> 00:04:11,288 partners I spoke to who I could no longer help or I couldn't help before, all of a sudden I was in position to go back to them and tell them the time is now. 58 00:04:11,288 --> 00:04:13,610 The technology has advanced to that point. 59 00:04:13,610 --> 00:04:15,730 And so I was ready to go all in. 60 00:04:15,730 --> 00:04:22,198 think others, this is back in 2023, 2024 timeframe, not everyone was so sold on the technology. 61 00:04:22,198 --> 00:04:28,670 think there was a, especially at that time, the initial knee jerk reaction of a lot of clients was like, don't use gen AI. 62 00:04:28,925 --> 00:04:32,016 There's a lot of fear, I think, early on, whereas I was leaning in. 63 00:04:32,016 --> 00:04:40,910 And so it just became clear that if I wanted to have the impact, a broad impact, that the moment demanded that it would be outside the firm at that point. 64 00:04:41,062 --> 00:04:41,612 Interesting. 65 00:04:41,612 --> 00:04:52,237 Yeah, and your days inside the firm, you had a big part in the ExtraNet that Morgan Lewis kind of built their own custom ExtraNet platform on SharePoint. 66 00:04:52,238 --> 00:04:52,778 Correct. 67 00:04:52,778 --> 00:04:53,138 Yeah. 68 00:04:53,138 --> 00:04:55,839 Was that SharePoint on-prem or SharePoint online? 69 00:04:57,160 --> 00:04:57,694 Okay. 70 00:04:57,694 --> 00:05:00,040 I think we eventually moved to SharePoint Online. 71 00:05:00,040 --> 00:05:00,800 Yeah. 72 00:05:00,800 --> 00:05:05,060 So, and that's kind of how you and I got connected was your experience around that. 73 00:05:05,060 --> 00:05:13,560 I've spent the last 17 years in the legal collaboration space started in 2008 with a company called Acrowire. 74 00:05:13,560 --> 00:05:21,660 And we did a lot of time and materials consulting, building custom solutions like the one that you guys built in-house. 75 00:05:21,660 --> 00:05:29,376 Most firms don't have those have the chops, you know, or the personnel to deliver on an engagement like that. 76 00:05:29,539 --> 00:05:40,814 Morgan Lewis being a big firm obviously did, but yeah, you learned a ton in that process and then your current role, you've kind of translated that and helped them with their 77 00:05:40,814 --> 00:05:42,445 portal product, correct? 78 00:05:42,696 --> 00:05:43,388 That's right, yeah. 79 00:05:43,388 --> 00:05:51,011 Tell us a little bit about kind of what the thinking was with Portal and where you see it going. 80 00:05:51,389 --> 00:06:00,849 Yeah, so the work was, you know, from the early days was envisioned to be a collaborative AI workspace for lawyers where you could collaborate initially internally within a firm. 81 00:06:00,849 --> 00:06:06,089 So you can share it between different teams who work in a firm, share content more broadly. 82 00:06:06,089 --> 00:06:16,189 As we were working with firms to develop work, you know, they would create workflows that came out earlier this year, the ability to kind of automate multiple multi-step workflows 83 00:06:16,189 --> 00:06:20,085 using AI and these fully agentic experiences. 84 00:06:20,085 --> 00:06:28,691 As we were building that out for uh our clients, for our customers, they uh began asking, this is really cool, this is really awesome that they have these experiences. 85 00:06:28,691 --> 00:06:31,974 And they were wondering if they could share that with their clients directly. 86 00:06:31,974 --> 00:06:35,216 broadly speaking, the way we see it is a couple of fronts. 87 00:06:35,216 --> 00:06:39,339 One is that AI is changing the speed of delivery. 88 00:06:39,339 --> 00:06:45,595 So the responsiveness that a firm can be now is far greater than it ever could have been before. 89 00:06:45,595 --> 00:06:57,022 the time of performing certain types of legal tasks, particularly things like first pass reviews or things like due diligence, these kinds of tasks are being accelerated greatly. 90 00:06:57,022 --> 00:07:05,216 And as you're able to complete these tasks a lot faster, clients expect faster, more responsive delivery of those services. 91 00:07:05,216 --> 00:07:09,609 And something like email, which is really what most of us are competing against, right? 92 00:07:09,609 --> 00:07:13,413 So email is by far the dominant way. 93 00:07:13,413 --> 00:07:16,026 that lawyers and clients collaborate today. 94 00:07:16,026 --> 00:07:22,561 Like that's the big competitor out there is uh just email and like kind of the status quo. 95 00:07:22,561 --> 00:07:28,725 And so that was maybe a suitable format when it took a month to do a due diligence. 96 00:07:28,725 --> 00:07:38,946 know, having the results come to you in a static Excel spreadsheet maybe was a suitably slow delivery mechanism and suitably, you know, advanced mechanism. 97 00:07:38,946 --> 00:07:43,446 for that kind of legacy work, work performed in a legacy manner. 98 00:07:43,546 --> 00:07:54,726 But once you can complete that work within a matter of days or within a day now, having it be performed by an AI and then double checked by attorneys, clients expect a quicker, a 99 00:07:54,726 --> 00:07:58,066 more enhanced service and delivery of that work product. 100 00:07:58,066 --> 00:08:04,666 And so now you're delivering the work product directly through a portal where they can then interact with the work product. 101 00:08:04,666 --> 00:08:08,226 So they can then use AI to then ask questions of that work product. 102 00:08:08,226 --> 00:08:11,378 And so it just creates this overall better experience for the customer. 103 00:08:11,624 --> 00:08:19,270 Yeah, and you and I had a conversation about kind of where, where Lagora portal fits into the bigger picture. 104 00:08:19,270 --> 00:08:28,525 And I think it, the, the capabilities that Lagora portal provides are a very narrow, but important slice of the big picture. 105 00:08:28,525 --> 00:08:30,256 I shared a graphic with you. 106 00:08:30,256 --> 00:08:31,987 I can share it quickly on the screen. 107 00:08:31,987 --> 00:08:33,528 Most people listen to this podcast. 108 00:08:33,528 --> 00:08:36,220 They don't actually, I might not be able to let's see here. 109 00:08:36,343 --> 00:08:49,654 just to kind of put in context where I see legora portal playing, which is really in that kind of practice space where, you know, intake document, Q and a playbooks, co-authoring, 110 00:08:49,735 --> 00:08:55,240 drafting, shared tabular review, all those sorts of things are very practice of law specific. 111 00:08:55,240 --> 00:09:00,445 And I see that as a necessary, but still fairly narrow slice. 112 00:09:00,445 --> 00:09:02,566 I expect that slice to get bigger. 113 00:09:02,619 --> 00:09:12,227 Over time as AI's capabilities grow and more work from the practice gets fed, becomes tech enabled. 114 00:09:12,227 --> 00:09:23,255 Um, but you know, when we were at a TLTF summit, you guys announced portal, like, I don't know, two or three days and everybody's like, Hey, are you worried about, you know, ex 115 00:09:23,255 --> 00:09:24,236 your extra net product? 116 00:09:24,236 --> 00:09:27,318 And I was like, no, these things are really complimentary. 117 00:09:27,359 --> 00:09:32,082 If anything, and I wrote an article about it on LinkedIn, I'm actually excited. 118 00:09:32,146 --> 00:09:45,747 that you guys and then Harvey later announced shared spaces, I'm excited that there's real money being put into this collaboration space because it's been so badly neglected. 119 00:09:45,747 --> 00:09:50,792 And you know this because you helped build a tool to facilitate it. 120 00:09:50,792 --> 00:09:59,595 It's been so badly neglected that there's very little overlap with what we do, really very little to anything in the practice of law space. 121 00:09:59,701 --> 00:10:11,628 we're more focused on the broader relationship, know, like open and closed matters, historical, you know, legal spend, attorney profile information that are billing time to 122 00:10:11,628 --> 00:10:18,081 the matters, the legal team, all of that sort of stuff, model document sharing, more kind of business of law stuff. 123 00:10:18,201 --> 00:10:25,704 So yeah, I was actually excited, but it really, the reason I wrote that article, I was shocked on how many people 124 00:10:25,750 --> 00:10:29,744 got confused by thinking that what you were doing competed with us. 125 00:10:29,744 --> 00:10:34,067 And these were by, I had a couple of former high Q guys asked me this. 126 00:10:34,067 --> 00:10:37,760 And I'm like, dude, you know what, you understand the landscape. 127 00:10:37,760 --> 00:10:40,191 I'm surprised that you think so. 128 00:10:40,291 --> 00:10:49,028 Anyway, we're glad to see you guys come in and actually in Harvey too, and put some investment here because it's been a badly neglected space. 129 00:10:49,028 --> 00:10:54,383 Yeah, I think that, you know, especially as somebody who, I mean, I first got involved in extranets really early in my career. 130 00:10:54,383 --> 00:11:01,747 And, you know, I certainly I had an impact where, you I own practice using them and delivering, you know, providing client reporting through these, through these portals. 131 00:11:01,747 --> 00:11:05,599 So the clients can understand what we were doing, like with the results that we were getting for our clients. 132 00:11:05,599 --> 00:11:09,071 So like communicating to them the substantive results of our practice. 133 00:11:09,071 --> 00:11:12,346 And certainly, you know, I think that the 134 00:11:12,346 --> 00:11:18,990 The adoption though of portals or extranets as know, oftentimes they're called has been fairly limited, right? 135 00:11:18,990 --> 00:11:25,343 It's been, you I mean, basically a little bit more advanced than a secure file transfer in a lot of instances. 136 00:11:25,343 --> 00:11:36,299 And so the potential, I think a lot of people have seen for a long time, myself included, like we saw robust potential in these client collaboration portals and enhancing the 137 00:11:36,299 --> 00:11:39,472 collaboration between uh law firms and their in-house counsel. 138 00:11:39,472 --> 00:11:41,543 But it's always, it's never lived up. 139 00:11:41,543 --> 00:11:42,893 to that potential. 140 00:11:42,893 --> 00:11:47,855 And I think that, even the things I did with Laura Lewis, just, we barely were scratching the surface. 141 00:11:47,855 --> 00:11:54,528 So I think that when I look at that pie chart you presented, I think it's probably more or less accurate as to where portals are today. 142 00:11:54,528 --> 00:12:03,621 I think that you're absolutely right that like portals today are not delivering, people really aren't delivering a lot of legal services, a lot of final work products. 143 00:12:03,621 --> 00:12:06,411 Certainly they're not creating new service lines, right? 144 00:12:06,411 --> 00:12:08,142 Like self-service legal service. 145 00:12:08,142 --> 00:12:08,690 uh 146 00:12:08,690 --> 00:12:11,570 service delivery through portal today. 147 00:12:11,570 --> 00:12:15,970 Very little of that's being done, particularly in the AMLOS space. 148 00:12:16,330 --> 00:12:17,990 So there's so much more upside. 149 00:12:17,990 --> 00:12:19,310 I think we're going to see a couple of things. 150 00:12:19,310 --> 00:12:21,890 One is you're going to see that pie grow. 151 00:12:21,890 --> 00:12:28,690 The demand for portal services, both on the business law and the practice of law side, is going to skyrocket. 152 00:12:28,690 --> 00:12:30,790 Again, just because we're so early days. 153 00:12:30,790 --> 00:12:36,658 And then I think also, just because practices of law has been such a small portion of what that is historically, 154 00:12:36,658 --> 00:12:37,718 if that's going to grow. 155 00:12:37,718 --> 00:12:39,778 And I think AI is going be the thing that drives that, right? 156 00:12:39,778 --> 00:12:44,938 Like I said earlier, it's going to be one, just you expect to see better experiences. 157 00:12:44,938 --> 00:12:54,489 know, like email is no longer suitable in an age of AI, clients know that their firms are using AI throughout the process to accelerate delivery. 158 00:12:54,489 --> 00:13:00,389 And then as firms are going the next, I think the next step though, is as firms are productizing legal services, right? 159 00:13:00,389 --> 00:13:01,801 So this is kind of the big... 160 00:13:01,801 --> 00:13:07,886 conversation I'm having the most with like management committees and managing partners or firms is uh what's on the horizon. 161 00:13:07,886 --> 00:13:17,573 You the idea that AI is going to disrupt certain things, you know, that it's going to change, you you won't be able to build like, know, when you can take that due diligence 162 00:13:17,573 --> 00:13:22,938 from 30 hours to like, to, you know, one or two, you can't charge 30 hours anymore, right? 163 00:13:22,938 --> 00:13:26,904 Like you can maybe charge two or better yet you reprice that work, right? 164 00:13:26,904 --> 00:13:30,867 You say, Hey, it's no longer be doing, you know, the value of the work. 165 00:13:31,027 --> 00:13:39,134 Maybe that correlates to the 30 hours, but now that it's two hours of attorney time, the value to the client is still substantial for that work. 166 00:13:39,134 --> 00:13:41,226 And it's not really reflected by those two hours. 167 00:13:41,226 --> 00:13:45,218 The metric has gone out of sync for the value that's being delivered to the client. 168 00:13:45,218 --> 00:13:46,909 And so you then want to reprice that work. 169 00:13:46,909 --> 00:13:51,373 Also, it doesn't account for the fact that you had to procure legal technology. 170 00:13:51,373 --> 00:13:56,327 It didn't account for the time and materials to build out the AI workflows that you're using. 171 00:13:56,327 --> 00:13:59,020 And so you need to capture that value somehow. 172 00:13:59,020 --> 00:14:07,811 And so if you reprice that work on a fixed fee basis or on a subscription fee basis, you're then able to, you you productize these things. 173 00:14:07,811 --> 00:14:17,871 And so as firms are seeing kind of what's on the horizon, what's really inspiring to me is that a lot of managing companies are asking like, how can we self disrupt, right? 174 00:14:17,871 --> 00:14:26,391 Like they don't want to be caught off guard and wake up one day when the phone stops ringing and realize the phone has stopped ringing or wake up one day and realize that 175 00:14:26,391 --> 00:14:27,435 they've been undercut. 176 00:14:27,435 --> 00:14:33,915 you know, on an RFP because their competitors are offering stuff on a fixed basis and they haven't done the groundwork, right? 177 00:14:33,915 --> 00:14:38,095 They haven't learned the technology, they haven't built out their workflows, they haven't productized their legal services. 178 00:14:38,095 --> 00:14:40,175 And so they're like caught off guard. 179 00:14:40,175 --> 00:14:42,275 And so a lot of them are being proactive. 180 00:14:42,275 --> 00:14:51,095 Now, obviously, I'm talking to more proactive firms, you know, the ones who have their heads completely in the sand about this, you know, probably aren't taking my calls or 181 00:14:51,095 --> 00:14:53,055 aren't taking the phone to call me. 182 00:14:53,055 --> 00:14:56,395 So I am talking to more forward thinking firms. 183 00:14:56,395 --> 00:15:03,035 But a lot of them are going that next step and really asking from the top down, asking the management committee is thinking about this. 184 00:15:03,035 --> 00:15:11,635 And then they're asking their partners, like throughout the partnership, kind of on a partner by partner basis, what are you doing to rethink pricing, packaging, and delivery 185 00:15:11,635 --> 00:15:13,706 of legal services in light of AI? 186 00:15:13,706 --> 00:15:20,846 One thing that one firm that I was talking to did that was particularly interesting is asking that as part of their year-end partner evaluations. 187 00:15:20,846 --> 00:15:24,906 So at the end of every year, you have to do a self-evaluation as a partner or a questionnaire. 188 00:15:25,002 --> 00:15:29,393 And they'll ask questions like, what have you done for continuity, business continuity? 189 00:15:29,393 --> 00:15:37,298 Like what have you done this year to ensure that if you were to leave the firm or if you were to the lottery or something that retire, that other people will be able to continue 190 00:15:37,298 --> 00:15:39,018 to the torch forward. 191 00:15:39,018 --> 00:15:40,939 So that's kind of a typical question they would ask. 192 00:15:40,939 --> 00:15:43,781 And now they're asking, what have you done with AI, right? 193 00:15:43,781 --> 00:15:46,553 In order to rethink legal services pricing. 194 00:15:46,553 --> 00:15:49,134 And they're getting ahead of things and re-productizing that. 195 00:15:49,134 --> 00:15:51,540 And then the way, once you productize things, 196 00:15:51,540 --> 00:15:58,563 Meaning you take your knowledge and experience of your firm and you embed it into reusable workflows that are AI powered. 197 00:15:58,563 --> 00:16:01,204 You then have to deliver those some power. 198 00:16:01,204 --> 00:16:08,986 And you could just have it be, one option would be to have it just be something that's available to the associate and the associate pulls the lever and then emails the work 199 00:16:08,986 --> 00:16:09,666 product. 200 00:16:09,666 --> 00:16:14,868 But clients know that you're just pulling a lever and they don't want to really, it's not like, it's not a great experience. 201 00:16:14,868 --> 00:16:18,203 And there's just so much more we can do in terms of like interactive experiences. 202 00:16:18,203 --> 00:16:27,348 the ability to ask follow-up questions to able to provide due diligence in this tabular review, which you can do with Legora, which is this interactive spreadsheet-like 203 00:16:27,348 --> 00:16:32,251 experience where you can go into each cell and you can see what the AI's reasoning was. 204 00:16:32,251 --> 00:16:38,003 You can see an audit log of when the AI generated the answer, the associate who signed off on that answer. 205 00:16:38,003 --> 00:16:44,465 To be able to have that level of transparency into the work product is something that we weren't able to have before. 206 00:16:44,465 --> 00:16:52,630 And so yeah, as they're rethinking productizing, a lot of them are looking at the portal as a way to deliver these new legal services directly to their clients. 207 00:16:52,630 --> 00:17:03,922 Yeah, and you and I talked a little bit about like, why is it that such little sophistication has made its way into these client facing portals? 208 00:17:03,922 --> 00:17:07,025 And I think it's a combination of two things. 209 00:17:07,025 --> 00:17:09,127 I'm curious your take on this. 210 00:17:09,127 --> 00:17:14,271 I think that just inherently, law firms are reluctant to share. 211 00:17:14,351 --> 00:17:15,953 They don't want to overshare. 212 00:17:15,953 --> 00:17:17,942 And that's a very 213 00:17:18,000 --> 00:17:19,501 lawyer mindset, right? 214 00:17:19,501 --> 00:17:23,693 And it's a good one from a legal perspective, right? 215 00:17:24,173 --> 00:17:28,156 Additional information that's not essential can create liability, right? 216 00:17:28,156 --> 00:17:29,257 In a legal context. 217 00:17:29,257 --> 00:17:36,402 And so law firms have a, they have a natural stance of being conservative on what they share. 218 00:17:36,402 --> 00:17:39,464 And then clients haven't really demanded it. 219 00:17:39,464 --> 00:17:43,806 Like, you know, we compete primarily with high Q and the number of 220 00:17:43,904 --> 00:17:55,260 Like basic file sharing sites that we see, it's probably four or five to one in terms of anybody doing anything sophisticated with it. 221 00:17:55,260 --> 00:18:00,843 And you know, their architecture is very much kind of isolated and it made a lot of sense. 222 00:18:00,843 --> 00:18:09,368 Like when Haiki first came out, the cloud was big and scary to law firms, you know, so you really, they had to do a lot of work to ring fence and isolate. 223 00:18:09,368 --> 00:18:12,364 And that is really what allowed them to 224 00:18:12,364 --> 00:18:23,582 to grow so quickly, but now that firms are comfortable with the cloud and all the big AMLaw firms are either there or making their way there, that additional friction, that air 225 00:18:23,582 --> 00:18:30,067 gap, now makes it challenging when firms want to build integrated experiences. 226 00:18:30,067 --> 00:18:40,589 When we show a client relationship page, we deploy in the client's tenant, so we can tap into their practice management solution, their DMS. 227 00:18:40,589 --> 00:18:45,860 experience management, again, rich attorney profiles for everybody that's spending time to those matters. 228 00:18:45,860 --> 00:18:55,664 We can pull into their analytics and show law firms never share whip, legal spend over time, all of these sorts of things. 229 00:18:55,664 --> 00:19:00,766 the current model, the current incumbent in that space, the architecture makes that challenging. 230 00:19:00,766 --> 00:19:07,097 we see a lot of opportunity there to do things differently, but 231 00:19:07,099 --> 00:19:11,287 What is your take on why it's been just file sharing? 232 00:19:11,287 --> 00:19:13,982 it those dynamics or are there other ones? 233 00:19:13,982 --> 00:19:15,784 I think that's a good point. 234 00:19:15,784 --> 00:19:20,928 think that probably the biggest reason is that the attorneys themselves aren't in those systems, right? 235 00:19:20,928 --> 00:19:27,891 So, they're certainly going to be reluctant to grant access to a platform that they're not themselves actually using on a day-to-day basis. 236 00:19:27,992 --> 00:19:37,309 Because the protection, can't really be overstated how sacrosanct like the attorney-client relationship is to partners. 237 00:19:37,309 --> 00:19:41,580 how protective partners can be over those relationships. 238 00:19:41,580 --> 00:19:48,620 Because obviously they're the lifeblood of the firms, but they're also the lifeblood of the individual partners who were working on them. 239 00:19:48,700 --> 00:19:55,671 so ensuring that you don't, that you control the narrative is very important to clients. 240 00:19:55,671 --> 00:19:59,211 Like you said, work in progress, like showing them stuff like was incomplete. 241 00:19:59,211 --> 00:20:03,131 You don't want them to kind of see your internal impressions, right? 242 00:20:03,131 --> 00:20:04,944 Like you might have an associate who, 243 00:20:04,944 --> 00:20:09,705 was working on a draft and you wouldn't want to show like, and they asked, they're like, is this really what we want? 244 00:20:09,705 --> 00:20:10,966 Or what about this other counter argument? 245 00:20:10,966 --> 00:20:15,527 And that might not be the narrative that you ultimately want to present to the client. 246 00:20:15,527 --> 00:20:24,246 And so it can cause uncertainty in your analysis if the, know, if you inadvertently share, you know, certain materials to, to a client. 247 00:20:24,246 --> 00:20:28,413 So I think there's been, there's a lot of protection around that relationship. 248 00:20:28,413 --> 00:20:33,028 And so when you add to the fact that like the, the systems you're mentioning of integrating, 249 00:20:33,028 --> 00:20:36,374 and being able to display, the partners themselves aren't really in those platforms. 250 00:20:36,374 --> 00:20:43,320 I think what's changed, least for Legora, is that we actually have partners and associates like you're in Legora daily, right? 251 00:20:43,320 --> 00:20:53,130 And so we're not coming from the outside and going to a partner and saying, hey, we want to share your content, your work product with your clients, and you're not using it. 252 00:20:53,130 --> 00:20:57,894 No, it's the partners and the associates who are using Legora who are saying, hey, 253 00:20:57,894 --> 00:21:03,227 we have this work product, is there a way that we can control how we share this with our clients? 254 00:21:03,227 --> 00:21:12,322 And so, and I would distinguish the kind of two use cases of, we have a final work product and we want to share that with the client and introduce them to the workspace versus like 255 00:21:12,322 --> 00:21:21,089 the more self-service productized legal services themselves, as they're rethinking the service delivery, making that obviously if it's self-service, you need some place where 256 00:21:21,089 --> 00:21:25,569 the clients are going to come in to perform the legal services or to have it be performed. 257 00:21:25,569 --> 00:21:31,593 ah But in both instances, it's the attorneys, partners, the associates are already in the platform using it. 258 00:21:31,593 --> 00:21:33,805 And then they want to introduce clients to that environment. 259 00:21:33,805 --> 00:21:35,477 So they're already comfortable with the environment. 260 00:21:35,477 --> 00:21:44,003 They just want to be able to know exactly, and this is from a product design perspective, what's really important, you know, to make sure that as we're building this product, that 261 00:21:44,003 --> 00:21:49,717 it is absolutely crystal clear what's being shared with the client at all times so that the partners feel comfortable. 262 00:21:49,717 --> 00:21:50,928 They're already comfortable in the environment. 263 00:21:50,928 --> 00:21:54,538 Now they're comfortable with, okay, they know exactly what they're sharing with whom. 264 00:21:54,538 --> 00:21:55,688 and what the clients can see. 265 00:21:55,688 --> 00:22:03,122 And so, you we allow them, for example, if you're sharing a workflow with clients, you can have them just see the output, but not see the reasoning. 266 00:22:03,122 --> 00:22:06,283 You can have them see the reasoning and the output, but not the prompts. 267 00:22:06,283 --> 00:22:06,623 Right? 268 00:22:06,623 --> 00:22:15,907 So we allow that granular control, which is particularly important as they productize their legal services, because they're basically taking their hundreds of years, know, or 269 00:22:15,907 --> 00:22:19,098 decades of knowledge and experience they have representing. 270 00:22:19,098 --> 00:22:23,330 mean, there's a reason why top tier firms command a premium in the market. 271 00:22:23,330 --> 00:22:27,954 um And one of those is that the cumulative experience they have handling a certain type of work. 272 00:22:27,954 --> 00:22:32,958 And so they're taking that and they're embedding that into technology, into these workflows. 273 00:22:32,958 --> 00:22:35,970 And so they don't, that becomes their IP, right? 274 00:22:35,970 --> 00:22:44,006 They're something that was in somebody's head and they're embedding it into a workflow and it is now firm IP that the firm wants to protect and needs to protect. 275 00:22:44,006 --> 00:22:50,018 And so to be able to ensure that as they're sharing the workflows, the work can get done, client can come in and say, 276 00:22:50,018 --> 00:22:59,228 run a contract playbook through the Legora portal, the client can have the playbook be run and they can get the output from uh Legora, but they don't necessarily see the playbook 277 00:22:59,228 --> 00:23:01,590 itself that is the IP of the firm. 278 00:23:01,590 --> 00:23:03,041 Yeah, that makes sense. 279 00:23:03,041 --> 00:23:05,813 Like how did, how should customers think about. 280 00:23:05,813 --> 00:23:16,229 Legora clients or other vendors clients when they make systems client facing and they decide to pivot at some point, right? 281 00:23:16,229 --> 00:23:21,482 Like they, they want, is there operational risk that they need to think about? 282 00:23:21,482 --> 00:23:27,946 Like, what if you, like, let's say you're a Legora client today and you share. 283 00:23:28,038 --> 00:23:38,593 through Lagora portal, a client really likes it and let's say the partner wants to switch to another platform or like how do you get that kind of pushback? 284 00:23:38,593 --> 00:23:42,665 Like, hey, if I show this to my clients and I'm kind of, we're getting hitched here. 285 00:23:42,665 --> 00:23:45,568 Do you ever get, have those sorts of conversations? 286 00:23:45,568 --> 00:23:48,270 Yeah, I mean, I guess it's an interesting question. 287 00:23:48,270 --> 00:23:53,882 So I think it highlights the importance of vendor selection, And like vendor durability. 288 00:23:53,882 --> 00:23:58,460 think one question that does come up in the legal tech space right now, there's a lot of players. 289 00:23:58,460 --> 00:24:04,565 uh It's every year at Iltta or Legal Week, there's tons of new players in the market. 290 00:24:04,565 --> 00:24:12,588 And with an industry that's happening so fast, everything's happening so fast and there's so many new market entrants in terms of new legal tech companies popping up all the time. 291 00:24:12,588 --> 00:24:16,411 And we're now starting to see some of them disappear. 292 00:24:16,411 --> 00:24:26,057 it causes a concern, obviously, on the part of firms, as well as their clients, about vendor durability and who they want to work with. 293 00:24:26,057 --> 00:24:34,893 particularly, where we've been thus far, think, been quite flexible, I think, in terms of like, if you're talking about just like a basic AI legal assistant, there's not a lot of 294 00:24:34,893 --> 00:24:36,834 risk or switching costs there. 295 00:24:37,052 --> 00:24:47,517 But as we really look to the future, know, as firms look to embed their knowledge experience into workflows, they have to be, it's incredibly important that they go with 296 00:24:47,517 --> 00:24:49,377 vendors that they trust. 297 00:24:49,377 --> 00:24:55,601 mean, we're basically asking customers to help us build the future of legal practice together. 298 00:24:55,601 --> 00:25:00,904 And as we do that, you know, it's very important who they select to build that future. 299 00:25:00,948 --> 00:25:06,651 And obviously there are a couple of things that clients or customers look to when deciding among vendors. 300 00:25:06,651 --> 00:25:08,532 I think financing is one of them. 301 00:25:08,532 --> 00:25:13,636 You want somebody who has sufficient uh capital to stand the test of time. 302 00:25:13,636 --> 00:25:16,919 Lawyers are notoriously peer obsessed. 303 00:25:16,919 --> 00:25:24,324 So looking to their peers and seeing what other firms are going with, gives them lot of confidence in terms of moving forward. 304 00:25:24,324 --> 00:25:26,513 But I think that the answer really is to just 305 00:25:26,513 --> 00:25:28,776 We're going to build the future together with these firms. 306 00:25:28,776 --> 00:25:32,139 And ultimately, if they're successful, we're going to be successful. 307 00:25:32,139 --> 00:25:38,165 And the hope would be is that they're successful, there's not going to be a need for them to move uh to another platform. 308 00:25:38,165 --> 00:25:43,389 What they're hoping to achieve out of this, in a lot of instances, is client stickiness. 309 00:25:43,389 --> 00:25:51,879 So they're hoping to entrench their client relationships via these tools and via these technologies. 310 00:25:51,879 --> 00:26:02,919 If they are the go-to platform for, like if you create like a self-service labor and employment portal for a client, and the client in-house counsel is using it daily to 311 00:26:02,919 --> 00:26:08,990 assess HR, to do compliance, to do customer investigations, to do reductions in forces, right? 312 00:26:08,990 --> 00:26:16,910 Like all these different aspects of labor and employment practice, if that can all be channeled through something like the Legora portal, you're going to create incredible 313 00:26:16,910 --> 00:26:19,646 client stickiness with that platform. 314 00:26:19,646 --> 00:26:24,606 And you wouldn't want to move at that point. 315 00:26:24,624 --> 00:26:24,954 Yeah. 316 00:26:24,954 --> 00:26:35,488 And you you had experience when you were on the law firm side with kind of like, it's interesting how many, so we don't have a ton of competition, thankfully, you know, on the 317 00:26:35,488 --> 00:26:41,231 intranet side, we had one competitor that's really kind of gone from the marketplace for the most part. 318 00:26:41,231 --> 00:26:49,296 And we're the only, it's either us or custom dev on the, on the extranet side, there's an incumbent in that space, which I talked about earlier. 319 00:26:49,296 --> 00:26:52,877 and you know, it's really, I see customers. 320 00:26:52,877 --> 00:26:55,919 debating, we buy it or build it? 321 00:26:56,039 --> 00:27:08,859 And it's such an easy, like we spent 14 years as Acrowire building with law firms and the ability for us to take what we built and hand over to them where they could manage the 322 00:27:08,859 --> 00:27:18,415 care and feeding and the enhancements and the support and the training and keeping their ear out, ear to the ground for any integrations that would implement breaking changes. 323 00:27:18,415 --> 00:27:21,237 Like they're just, it's such a high bar. 324 00:27:21,385 --> 00:27:28,337 And a lot of times law firms will go, but it's only going to cost me, you know, 600,000 to build this thing out. 325 00:27:28,337 --> 00:27:32,708 And I'm like, yeah, that's, that's the part that has a discrete beginning and end. 326 00:27:32,708 --> 00:27:37,070 It's the ongoing costs that really drive TCO. 327 00:27:37,070 --> 00:27:41,282 sometimes four to five X your initial development costs. 328 00:27:41,282 --> 00:27:50,494 So like, it blows my mind that I still have to have those conversations, but the reality is I do, because some, some firms have 329 00:27:50,540 --> 00:27:52,460 dev teams and guess what? 330 00:27:52,460 --> 00:27:54,380 Devs love to dev, right? 331 00:27:54,380 --> 00:27:54,860 I did. 332 00:27:54,860 --> 00:27:56,380 I used to be a developer. 333 00:27:56,520 --> 00:27:59,780 Um, and you know, it's, that's what you want to do. 334 00:27:59,780 --> 00:28:09,880 You want to flex those muscles, but like, I don't know, do you run into, how, how do you, know, in your experience, not just today with the Gora, but like, how have you thought 335 00:28:09,880 --> 00:28:13,351 about that problem kind of build versus buy in the law firm context? 336 00:28:13,351 --> 00:28:22,052 Yeah, so, you know, when I started, you know, really 2013, like started working with firm IT, you know, the legal tech market was not as mature as it is today. 337 00:28:22,052 --> 00:28:25,312 So I mean, guys, InfoDash was not around at that time. 338 00:28:25,312 --> 00:28:34,472 You know, so the, if I were to start down this journey today, or if there were an associate at a firm who were to start down the journey today where I was, like, there's no 339 00:28:34,472 --> 00:28:37,832 world in which they would have built the software internally. 340 00:28:37,984 --> 00:28:40,224 Not to say you wouldn't do some internal development. 341 00:28:40,224 --> 00:28:44,504 I think there's absolutely a role for certain types of development within firms, even today. 342 00:28:44,504 --> 00:28:48,904 But you certainly wouldn't have built out the matter management platform as I did. 343 00:28:48,904 --> 00:28:58,204 What we would have done instead is we would have got one of the top tier matter management type platforms that exists on the market that are fully customizable. 344 00:28:58,204 --> 00:29:01,784 We would have bought that or licensed that and then we would have just customized. 345 00:29:02,124 --> 00:29:06,095 If I were to advise somebody today, that's the route you go down. 346 00:29:06,095 --> 00:29:11,495 build a fully customizable, you don't internally build a customizable amount of management platform yourself. 347 00:29:11,955 --> 00:29:16,655 Because you could, mean, you could, but it's it's not, there's just too many options now. 348 00:29:16,655 --> 00:29:20,435 So the calculus has changed from a decade ago. 349 00:29:20,435 --> 00:29:23,115 Now there's lots of great solutions to the market. 350 00:29:23,135 --> 00:29:27,655 Certainly, as I said, I did a client, I did the extra platform at Morgan Lewis. 351 00:29:27,655 --> 00:29:30,595 And today it wouldn't make sense to do that. 352 00:29:30,595 --> 00:29:32,835 InfoDash wasn't in the extra net space at that time, right? 353 00:29:32,835 --> 00:29:35,568 So it wouldn't make sense to do that today. 354 00:29:35,568 --> 00:29:39,212 I do, you know, it's interesting to say devs like to dev, that's absolutely true. 355 00:29:39,212 --> 00:29:42,356 And the flip side of that is they don't really like to do maintenance and support. 356 00:29:42,356 --> 00:29:50,103 And so, that's, they want to work on the new thing and they don't want to, you know, uh tend to the legacy stuff. 357 00:29:50,103 --> 00:29:54,366 So that creates a, you know, certainly like it increases the total cost of ownership of this, these tools. 358 00:29:54,366 --> 00:29:58,030 And of course, if you built something fully custom, you're almost wedded to the people who built it, right? 359 00:29:58,030 --> 00:30:00,350 Because those people are the only ones who know. 360 00:30:00,350 --> 00:30:02,221 how that system operates. 361 00:30:02,221 --> 00:30:06,764 And it's difficult to, uh you can't just go off the street and find somebody who knows how to do it. 362 00:30:06,764 --> 00:30:13,607 Whereas like, you if you were to use, just take a non-legal example, if you were to build like a custom solution on top of Salesforce, you know, there's a whole marketplace of 363 00:30:13,607 --> 00:30:19,433 people out there who have Salesforce dev skills, who could support your application if, you know, somebody were to leave your company. 364 00:30:19,433 --> 00:30:23,916 So there's certainly lots of downstream implications of doing custom dev. 365 00:30:23,916 --> 00:30:26,057 The stuff that does make sense, 366 00:30:26,064 --> 00:30:36,437 you know, in terms of custom development in firms, integration work makes a lot of sense given that uh each firm's systems, know, even if they use like say one of the leading, you 367 00:30:36,437 --> 00:30:44,551 know, timing and billing systems or, you know, the, each one is so customized to each firm, you know, the amount of complexity there is something that most people, unless 368 00:30:44,551 --> 00:30:52,439 you've been kind of on the database side of things, you don't know how complicated like timing and billing systems are really, you know, like the matter of practice management 369 00:30:52,439 --> 00:30:54,095 systems inside of firms. 370 00:30:54,095 --> 00:30:56,578 each one is highly bespoke to that firm. 371 00:30:56,578 --> 00:31:06,816 And so the work that you need to do to say integrate a new business intake platform with another like matter management system is highly specific, is not something you're going to 372 00:31:06,816 --> 00:31:07,938 be able to find off the shelf. 373 00:31:07,938 --> 00:31:11,030 So that kind of work really makes sense. 374 00:31:11,030 --> 00:31:20,436 We do have the conversation kind of today in my role where I'm talking with firms all the time and I'll show them something you can do with Legar and they'll say, 375 00:31:20,496 --> 00:31:21,996 we built that. 376 00:31:21,996 --> 00:31:24,296 then of course I asked them, like, well, how long did it take you to build that? 377 00:31:24,296 --> 00:31:27,216 And they'll say, oh, well, it took us like four months and we did it like a year ago. 378 00:31:27,216 --> 00:31:28,596 And it's like, awesome. 379 00:31:28,596 --> 00:31:34,736 know, that, you probably that was something that Lagora didn't do a year ago, but you know, today I did it in Lagora in an hour. 380 00:31:34,896 --> 00:31:42,816 And so if once things become commercially available, right, like once you can use commercial software and you can just take off the shelf commercial software and customize 381 00:31:42,816 --> 00:31:48,936 it to meet your use case, it no longer makes sense to dedicate development resources internally to reinvent the wheel. 382 00:31:49,460 --> 00:31:59,610 Instead, it makes sense to dedicate those resources to things that are highly practice or matter specific that uh you wouldn't be able to find commercially available. 383 00:31:59,610 --> 00:32:07,167 things that, uh you know, and then it looks, makes sense to integrate those with a platform like Legora via something like MCP, right? 384 00:32:07,167 --> 00:32:14,696 So that way we can handle sort of commercially available use cases, document review, document traction, things like word add-in. 385 00:32:14,696 --> 00:32:22,292 we can build things like great user experiences, interfaces, permissioning and security and all those tasks that aren't necessarily the most exciting. 386 00:32:22,292 --> 00:32:27,155 And then firms can focus on what sort of is their competitive advantage, which is their knowledge and experience, right? 387 00:32:27,155 --> 00:32:36,389 And so they can focus on building technology that really lends to their own strengths and let us at, in the tech vendor space, let us build great software. 388 00:32:36,435 --> 00:32:37,206 Yeah. 389 00:32:37,206 --> 00:32:38,506 Yeah, that makes sense. 390 00:32:38,506 --> 00:32:45,592 How do you think firms, what is the differentiation equation going to look like, you know, five years from now? 391 00:32:45,592 --> 00:32:45,932 Right? 392 00:32:45,932 --> 00:32:52,436 Like you can't buy an off the shelf solution and differentiate yourself based on that tool. 393 00:32:52,436 --> 00:33:02,925 You really kind of have to, you know, I have a theory about this, that law firms are going to, their special sauce is all of the knowledge and documents that created winning 394 00:33:02,925 --> 00:33:04,776 outcomes for their clients. 395 00:33:04,872 --> 00:33:05,132 Right. 396 00:33:05,132 --> 00:33:08,104 And how do they use that in a wholesale way today? 397 00:33:08,104 --> 00:33:11,305 Like, Legora is a, is a hosted solution, right? 398 00:33:11,305 --> 00:33:17,589 You're kind of bringing your data to AI versus bringing AI to your data. 399 00:33:17,589 --> 00:33:22,162 Like how, how are firms, how are firms going to differentiate themselves? 400 00:33:22,162 --> 00:33:25,194 Cause again, off the shelf tools by themselves aren't going to do it. 401 00:33:25,194 --> 00:33:32,752 Maybe it's through the workflows that you're talking about, but yeah, how is firm A going to differentiate from firm B? 402 00:33:32,752 --> 00:33:34,032 three to five years from now. 403 00:33:34,032 --> 00:33:38,184 Yeah, so I think what you're getting at though is that there's kind of a short and a long term. 404 00:33:38,184 --> 00:33:46,407 So in the short term, there is some early mover advantages of people who, you know, who lean into the tools, you know, not just licensing, but like who really get up to speed and 405 00:33:46,407 --> 00:33:48,988 integrate tools throughout their firms today, right? 406 00:33:48,988 --> 00:33:55,431 Because they have a strategic advantage over their competitors, say a trial, because they're using AI and their competitors aren't using AI. 407 00:33:55,431 --> 00:34:02,084 But certainly, uh once this kind of short term, once, you know, once everybody has a legora, for example, right? 408 00:34:02,150 --> 00:34:05,474 and it's become the incumbent solution or has some other tools. 409 00:34:05,474 --> 00:34:10,639 It won't be, you're right to say that you can't, there won't be a competitive advantage because everybody has it, right? 410 00:34:10,639 --> 00:34:18,467 I mean, so this will be sort of like, if we go back and kind of by analogy, you know, there were some competitive advantages early on with uh say legal research, electronic 411 00:34:18,467 --> 00:34:19,708 legal research, right? 412 00:34:19,708 --> 00:34:26,578 Early on, if you were a firm who leaned into electronic legal research, you were able to do more comprehensive legal research than your competitors were. 413 00:34:26,578 --> 00:34:35,058 that then gave you some strategic advantages at trial because you were able to find that one case that was on point or that case that rebuts what you're, know, that, that 414 00:34:35,058 --> 00:34:38,838 distinguished the precedent on which your opponent was relying. 415 00:34:38,838 --> 00:34:42,138 And so you're able to use, you gain a strategic advantage through that. 416 00:34:42,138 --> 00:34:46,498 But that then became very quickly because this is like a very competitive market. 417 00:34:46,578 --> 00:34:51,338 You know, once you're using it, once kind of the cat's out of the bag on electronic legal research, everybody's doing it. 418 00:34:51,338 --> 00:34:52,418 It just becomes the new normal. 419 00:34:52,418 --> 00:34:54,058 Like everybody has to do it. 420 00:34:54,074 --> 00:34:57,867 So there's no longer really any competitive advantage to using electronic research. 421 00:34:57,867 --> 00:34:59,199 It's just the baseline, right? 422 00:34:59,199 --> 00:35:06,185 And so in some ways, that will be the case with uh AI as well, is that like it will just become the new baseline. 423 00:35:06,185 --> 00:35:13,791 In the meantime, though, those firms that lean into it will, you know, do have a chance to get ahead and to take advantage of that, kind of seize the moment and get ahead. 424 00:35:13,791 --> 00:35:16,763 So I think that, you know, what will be their competitive advantage? 425 00:35:16,763 --> 00:35:19,795 mean, the good news in some ways is that it won't be the software, right? 426 00:35:19,795 --> 00:35:27,655 If they don't have to worry about like once everybody has AI, once they've, it's been the law schools are teaching it and all the associates know and the partners are using it on a 427 00:35:27,655 --> 00:35:28,915 day to day basis. 428 00:35:28,935 --> 00:35:32,095 It won't be the tech that you buy that's going to be your competitive advantage. 429 00:35:32,095 --> 00:35:36,034 It's going to go back to being the core things that you went to law school for, right? 430 00:35:36,034 --> 00:35:44,235 Look at the core things that firms use to differentiate today, which is like their people, their experience, their knowledge, their work product, their relationships, their brand, 431 00:35:44,235 --> 00:35:47,095 like those core things they use to differentiate themselves today. 432 00:35:47,095 --> 00:35:49,236 Those will be the enduring things. 433 00:35:49,236 --> 00:35:55,125 that remain once AI has kind of reshaped the landscape and however it's going to do it. 434 00:35:55,125 --> 00:35:57,607 Yeah. 435 00:35:57,607 --> 00:36:08,255 what about the amount of services that inside teams are going to work that once went to a law firm that they're going to manage internally? 436 00:36:08,255 --> 00:36:12,117 That is going to displace a certain amount of law firm revenue. 437 00:36:12,117 --> 00:36:14,349 And that's not a Lagora problem or a Harvey problem. 438 00:36:14,349 --> 00:36:15,850 That's the tech. 439 00:36:15,850 --> 00:36:17,241 The tech is going to do this. 440 00:36:17,241 --> 00:36:20,978 And law firms just need to come to grips with, I think that they're 441 00:36:20,978 --> 00:36:25,370 credible estimates that 30 % of associate hours today are gone. 442 00:36:25,370 --> 00:36:33,903 Based on where the tech is today, you can take 30 % of associate hours and essentially tech is going to be delivering on that. 443 00:36:33,903 --> 00:36:38,534 It doesn't matter if it's your platform or the platform down the street. 444 00:36:38,534 --> 00:36:44,197 Firms have to move up the value chain, which creates a lot of other challenges. 445 00:36:44,197 --> 00:36:47,038 Well, that's how you train lawyers. 446 00:36:47,038 --> 00:36:48,679 Is the grunt work? 447 00:36:48,679 --> 00:37:00,185 You throw them in a room with a box of documents and you review them and you have a partner parachute in and and Provide guidance and this this tech kind of blows that all up 448 00:37:00,185 --> 00:37:10,632 So, you know, it creates it creates challenges like okay the leverage model associate development You know potentially having a revenue impact 449 00:37:11,043 --> 00:37:20,358 Do you think that there's sufficient, because when I talk to some people, they point to Javan's paradox and go, there's a ton of unmet need and that's going to more than offset. 450 00:37:20,358 --> 00:37:25,041 And that could be true to a certain extent, but demand isn't infinite. 451 00:37:25,041 --> 00:37:33,905 So it's unclear at this point how far up the complexity spectrum AI is going to make its way and how fast. 452 00:37:33,905 --> 00:37:35,907 We don't know the answers to that. 453 00:37:35,907 --> 00:37:43,929 In fact, when you talk to credible sources at Frontier Labs today, they've kind of reached their limits on scaling, right? 454 00:37:43,929 --> 00:37:55,245 Like we've consumed all the publicly available data that's out there and just throwing more parameters at a model is starting to have diminishing returns. 455 00:37:55,245 --> 00:38:04,670 So there's a lot of talk about like us entering into a new age of research where, you know, scaling sucked the air out of the room for like the last four years. 456 00:38:04,826 --> 00:38:14,075 And there really wasn't an emphasis in resources placed on research and new approaches around engineering. 457 00:38:14,075 --> 00:38:17,619 Now with inference, we have a path to make things better. 458 00:38:17,619 --> 00:38:19,551 even there, feel like we're... 459 00:38:19,551 --> 00:38:21,809 Do you think we're gonna need a new... 460 00:38:21,809 --> 00:38:25,813 Are LLMs gonna be the answer in three to five years? 461 00:38:25,813 --> 00:38:27,044 Do you see the trajectory? 462 00:38:27,044 --> 00:38:29,115 And I know it's hard to say for sure, but... 463 00:38:29,173 --> 00:38:38,317 Just based on what you know, do you see LLMs as being the future state in three to five years or might it be something different? 464 00:38:38,769 --> 00:38:45,513 Yeah, I mean, it's probably beyond my, know, I'm not a researcher in the LLM space, so I don't know the AI space more broadly. 465 00:38:45,513 --> 00:38:50,176 So I am probably outside of my, you know, was kind of above my pay grade, so to speak. 466 00:38:50,176 --> 00:38:54,464 I mean, I would imagine it's hard for me to based on what I know in my experience of it. 467 00:38:54,464 --> 00:38:59,162 You know, there are people talk about world models being kind of the next frontier of AI. 468 00:38:59,162 --> 00:39:00,152 And I can see that. 469 00:39:00,152 --> 00:39:05,275 But I still kind of haven't gotten a very clean answer as why a why LLMs couldn't 470 00:39:05,275 --> 00:39:10,690 be the source data for creating synthetic data for that, to create scenarios that world models could then be trained on. 471 00:39:10,690 --> 00:39:16,473 So I would find it very perplexing if what we have today isn't a stepping stone to whatever's next. 472 00:39:16,473 --> 00:39:22,237 um Is a new breakthrough, a new paradigm going to be necessary to get to that new location? 473 00:39:22,237 --> 00:39:23,908 Yeah, to that new tier, likely. 474 00:39:23,908 --> 00:39:30,482 You're kind of referring to basically foundational models plateauing, the idea that scaling isn't going to get us more 475 00:39:30,482 --> 00:39:34,302 I think, yeah, it's quite possible. 476 00:39:34,382 --> 00:39:43,262 When I think about for legal though, I think that even if foundational models plateaued, we're barely scratching the surface, just in terms of just the number of attorneys who are 477 00:39:43,262 --> 00:39:43,722 using it. 478 00:39:43,722 --> 00:39:50,622 I I didn't see a stat to like half of all attorneys in firms of 500 or more are using AI last year. 479 00:39:50,622 --> 00:39:53,093 Okay, well, like they used it, what, once? 480 00:39:53,093 --> 00:39:55,769 The amount of adoption is still... 481 00:39:55,769 --> 00:39:57,180 uh relatively low. 482 00:39:57,180 --> 00:39:59,561 mean, obviously it's highly unequal. 483 00:39:59,561 --> 00:40:03,772 There are some people who using it every single day in some firms who have really high usage. 484 00:40:03,772 --> 00:40:13,199 But just industry-wide, uh we're still early days in terms of just total number of people who are using it and the value they're getting out of it. 485 00:40:13,199 --> 00:40:22,244 Okay, somebody might be using it to proofread an email, but that's really low-hanging fruit in terms of what these AI models today are capable of doing. 486 00:40:22,244 --> 00:40:24,914 So from a legal AI perspective, 487 00:40:24,914 --> 00:40:29,898 whether the models plateau or not, there's still plenty of value that we're going to be able to get. 488 00:40:29,898 --> 00:40:31,979 I don't think the models are going to plateau anytime soon. 489 00:40:31,979 --> 00:40:36,341 I do think there will be those breakthroughs. 490 00:40:36,341 --> 00:40:45,387 I don't have any specific insight into what those will be, but I think that given how much progress has been made, it'd be kind of silly to bet against them, to bet against them 491 00:40:45,387 --> 00:40:46,568 being a breakthrough. 492 00:40:46,568 --> 00:40:50,906 And we're making a civilizational bet here on AI, right? 493 00:40:50,906 --> 00:40:56,546 the amount of money that's being invested in AI is probably the largest investment we've ever made as a species. 494 00:40:56,646 --> 00:41:03,046 And so, you know, this is, and the payoff is potentially going to be the most important technology since, you know, the invention of writing. 495 00:41:03,366 --> 00:41:07,125 So it's, you know, really exciting. 496 00:41:07,125 --> 00:41:13,146 And I think that even if we don't hit AGI or whatnot, I think we're going to still see incredible things come out of this. 497 00:41:13,506 --> 00:41:20,822 It's just, you know, in the short term, like, you know, we, it seems slow, like from GBT four to five and then five to 5.1 and two, like, 498 00:41:20,822 --> 00:41:24,062 It seems small, but if you just zoom out, it's accelerating. 499 00:41:24,062 --> 00:41:26,716 It really is impressive, the rate of... 500 00:41:26,716 --> 00:41:29,708 We just get used to things so quickly, right? 501 00:41:29,708 --> 00:41:31,029 And so we're not really impressed. 502 00:41:31,029 --> 00:41:34,080 But you raised a number of other issues too. 503 00:41:34,080 --> 00:41:36,902 So I think Jevin's paradox, I think that's probably right. 504 00:41:36,902 --> 00:41:39,162 I agree with you, is certainly not infinite. 505 00:41:39,162 --> 00:41:47,019 I was reading an article the other day about radiology and there were predictions really even before GNI, like in terms of predictive coding of... 506 00:41:47,019 --> 00:41:47,763 uh 507 00:41:47,763 --> 00:41:51,483 or of traditional machine learning on x-rays. 508 00:41:51,483 --> 00:41:56,557 People were predicting this back in 2016 that there would be like, don't go to, don't study radiology. 509 00:41:56,557 --> 00:42:01,449 Like that field's gonna die because the computers are going to, uh you just plant all that. 510 00:42:01,449 --> 00:42:11,293 There's more demand today for radiology than there ever has before because simply as the cost goes down to examine an x-ray, the number of doctors who are then sending things off 511 00:42:11,293 --> 00:42:13,764 to get analyzed is up, right? 512 00:42:13,764 --> 00:42:15,247 So like there are now... 513 00:42:15,247 --> 00:42:18,367 greater demand for radiologists than there's ever been before. 514 00:42:18,367 --> 00:42:22,127 And radiological services are being demanded higher than they've ever been before. 515 00:42:22,287 --> 00:42:24,127 And will that continue forever? 516 00:42:24,127 --> 00:42:25,807 Well, probably not, because nothing will. 517 00:42:25,807 --> 00:42:28,927 But I think that it certainly will continue for a long time. 518 00:42:28,927 --> 00:42:38,127 So even in another domain, just as to benchmark what we're thinking in terms of legal, that's a very highly specialized task that people thought was going to be automated very 519 00:42:38,127 --> 00:42:38,567 quickly. 520 00:42:38,567 --> 00:42:43,681 And what it turned out there, which I think is very relevant to legal by analogy, is that 521 00:42:43,681 --> 00:42:49,053 The mistakes that humans make, it just so happens that they're the kind of things that AI is really good at catching. 522 00:42:49,053 --> 00:42:52,954 And the mistakes that AI makes, it turns out that humans are actually really good at catching. 523 00:42:52,954 --> 00:42:58,777 And so when you combine both humans plus AI, it's going to be better than either one on its own. 524 00:42:58,777 --> 00:43:04,419 There is so much demand for legal services that never gets met because attorneys are costly. 525 00:43:04,419 --> 00:43:08,716 Obviously, there are different segments of the legal market. 526 00:43:08,716 --> 00:43:17,816 But there are plenty of businesses where you never bother to call an attorney because even in Fortune 500 companies, because you're like, well, it's not going to be worth the point 527 00:43:17,816 --> 00:43:21,696 two or the point three that they're going to take or the couple of hours they're going to bill me. 528 00:43:21,796 --> 00:43:27,736 So people are just running the risk all the time by not calling lawyers and not consulting with lawyers. 529 00:43:27,736 --> 00:43:37,452 And so once you can lower the cost of getting advice on a given question, the number of times people are going to call their attorneys is going to go up. 530 00:43:37,452 --> 00:43:40,632 And that's just on the compliance advisory space. 531 00:43:40,732 --> 00:43:48,612 Once we go to the litigation space, it's kind of the game theory there is actually even more interesting because it's like, what does the world look like if everybody you were to 532 00:43:48,612 --> 00:43:53,132 give like an AI attorney, or if you were able to power all the new law school grads? 533 00:43:53,132 --> 00:43:57,332 Because one thing that's interesting is law school attendance is up at an all time high this year. 534 00:43:57,332 --> 00:44:00,432 The number of people who are looking to go to law school is at an all time high. 535 00:44:00,432 --> 00:44:04,292 So let's assume we have some of the largest law school graduating classes ever. 536 00:44:04,292 --> 00:44:05,812 What are those people going to do? 537 00:44:05,872 --> 00:44:13,199 Well, maybe with AI, they're going be able to go help meet the long tail of unmet needs of all the people out there who currently can't afford attorneys. 538 00:44:13,199 --> 00:44:21,037 Well, so now what does the world look like when all of a sudden everybody, or at least you've doubled the amount of people who are able to afford legal services? 539 00:44:21,037 --> 00:44:23,239 Well, now they're able to sue. 540 00:44:23,239 --> 00:44:27,002 When they wouldn't have been able to sue before, they're able to vindicate their rights in ways they weren't able to before. 541 00:44:27,002 --> 00:44:31,098 And that's then going to cause a massive increase in demand for defense attorneys. 542 00:44:31,098 --> 00:44:36,178 to then defend all the companies who are now being sued by all these new AI enabled attorneys. 543 00:44:36,318 --> 00:44:44,409 So a lot of these things, the game theory of it is that as demand on one side of the V increases, the demand will go up on the other side. 544 00:44:44,409 --> 00:44:51,309 So I don't think we're going to see, you say 30 % of associate hours can be replaced with AI. 545 00:44:51,709 --> 00:44:54,849 On a task level, that may be true. 546 00:44:54,909 --> 00:44:58,553 But it certainly doesn't necessarily follow that like they're 547 00:44:58,553 --> 00:45:04,813 going to you're going need a third of your associates or that you know that that time won't get repurposed. 548 00:45:04,813 --> 00:45:14,053 And you know, on the training thing, there's a myth, I think there certainly is some amount of work, you you know, it's like how I learned to be an attorney and how you know, 549 00:45:14,053 --> 00:45:17,353 most of the partners out there learn to be attorneys, like they did it through grinding. 550 00:45:17,353 --> 00:45:21,373 But like, if we're honest with ourselves, like a lot of that work wasn't productive, right? 551 00:45:21,373 --> 00:45:23,273 A lot of it was just a time sink. 552 00:45:23,273 --> 00:45:25,073 It was just a way to get a lot of hours. 553 00:45:25,073 --> 00:45:27,403 It was a way to generate leverage for a firm. 554 00:45:27,403 --> 00:45:31,016 And it wasn't necessarily something that really trained us. 555 00:45:31,016 --> 00:45:33,498 And in fact, oftentimes it inhibits you. 556 00:45:33,498 --> 00:45:38,903 you're so busy doing the grunt work that you don't even have time to like learn the higher level skills. 557 00:45:38,903 --> 00:45:41,145 And so it's actually slowing your progress. 558 00:45:41,145 --> 00:45:49,612 And so as AI can handle more of that underlying work, you're able to then focus on the higher level task, both, you know, even as you're learning. 559 00:45:49,612 --> 00:45:54,986 And I guess the other thing too is that, you know, how are we going to learn or train associates? 560 00:45:54,986 --> 00:45:56,117 Like if it's like, 561 00:45:56,229 --> 00:46:01,671 ah if it's helping on a lot of these lower-level tasks, mean, the AI is clearly going to be helping us learn too, right? 562 00:46:01,671 --> 00:46:06,212 AI done, deployed properly, explains its results, right? 563 00:46:06,212 --> 00:46:09,003 It explains where it's getting its analysis. 564 00:46:09,003 --> 00:46:13,075 so oftentimes when people ask me this, particularly in law firm context, they're like, well, how are we going to train associates? 565 00:46:13,075 --> 00:46:16,377 And it's like, well, it presumes you do a good job of training today. 566 00:46:16,377 --> 00:46:20,646 Training in law firms today is highly unequal. 567 00:46:20,646 --> 00:46:25,606 Not all associates get the same mentorship opportunities as others, right? 568 00:46:25,606 --> 00:46:31,506 It's like, you're really, you know, it's very unequal who gets access to the mentorship opportunities. 569 00:46:31,566 --> 00:46:43,006 And so if the AI can help us kind of bridge that gap and you level that playing field where everybody has access to AI in a firm and they can get, the AI can have infinite time 570 00:46:43,006 --> 00:46:49,126 and infinite patience to explain how to do a due diligence, how to analyze a contract, how to write a brief. 571 00:46:49,126 --> 00:46:57,746 why, you know, when you, when I was an associate, I would get red lines of my briefs back, but there's no explanations as to why I had made a mistake or why that paragraph needed to 572 00:46:57,746 --> 00:46:59,746 be cut down or moved around. 573 00:46:59,806 --> 00:47:08,266 But with something like LaGaurat today, like we explain every single edit and there's detailed explanations as to why LaGaurat recommends that edit be made. 574 00:47:08,266 --> 00:47:11,526 And that's far more time and attention than like any partner would ever have. 575 00:47:11,526 --> 00:47:18,714 So I really think the AI is going to be, you know, there might be some bumps along the road, but I think the AI is actually going to be used to 576 00:47:18,714 --> 00:47:21,697 help us train some of the best attorneys we've ever seen. 577 00:47:21,806 --> 00:47:22,376 Yeah. 578 00:47:22,376 --> 00:47:22,627 Yeah. 579 00:47:22,627 --> 00:47:31,113 I don't disagree that there are much more effective methods than the traditional, you know, legacy method that we've used to train attorneys. 580 00:47:31,113 --> 00:47:35,656 There are absolutely more efficient and effective ways to do it. 581 00:47:35,656 --> 00:47:38,309 So I think I completely agree with you there. 582 00:47:38,309 --> 00:47:41,092 well, this has been a fantastic conversation, Kyle. 583 00:47:41,092 --> 00:47:47,366 like I said, you and I been chewing the fat in the, on the, the legal collaboration space for quite some time. 584 00:47:47,366 --> 00:47:48,975 So it's, uh 585 00:47:48,975 --> 00:47:52,567 I really appreciate you spending a few minutes with us this afternoon. 586 00:47:52,731 --> 00:47:54,422 Yeah, like I said, was a lot of fun. 587 00:47:54,422 --> 00:47:54,943 Awesome. 588 00:47:54,943 --> 00:47:55,394 All right. 589 00:47:55,394 --> 00:47:57,177 Well, listen, have a good rest of your day. 590 00:47:57,177 --> 00:47:57,958 We'll chat again soon. 591 00:47:57,958 --> 00:47:59,571 All right. 592 00:47:59,571 --> 00:48:00,142 Sure. -->