1 00:00:01,793 --> 00:00:03,640 Al, good to see you this morning. 2 00:00:03,682 --> 00:00:05,055 Good to see you, Ted. 3 00:00:05,777 --> 00:00:09,950 So you and I talked about getting together on a podcast. 4 00:00:09,950 --> 00:00:19,237 think it was at evolve last year when we both won basically the same award, but you were on the law firm side. 5 00:00:19,237 --> 00:00:27,203 I was on the vendor side and I think we're still the reigning innovators of the year for a few more months. 6 00:00:29,926 --> 00:00:31,087 Exactly. 7 00:00:31,087 --> 00:00:32,399 So, um, 8 00:00:32,399 --> 00:00:33,360 Yeah. 9 00:00:33,360 --> 00:00:38,042 What's interesting is, you came from NRF, right? 10 00:00:38,042 --> 00:00:50,920 You were NRF Canada and we just made NRF US a client of InfoDash and now you're at Gowling, who's also a client of InfoDash, an early adopter and you guys are doing some 11 00:00:50,920 --> 00:00:51,971 really cool stuff. 12 00:00:51,971 --> 00:01:00,185 I hope you guys apply for, I don't know if you've had a chance to see the internet there, but y'all did an amazing job with the layout and design. 13 00:01:00,396 --> 00:01:04,443 Yeah, I just, actually had a demo this week, so really happy with it. 14 00:01:04,443 --> 00:01:06,504 Yeah, it looks really good. 15 00:01:07,765 --> 00:01:12,647 So yeah, before we jump into the agenda, let's get you introduced. 16 00:01:13,428 --> 00:01:22,192 You're now the National Director of AI, Innovation and Knowledge at Gowling. 17 00:01:22,433 --> 00:01:27,515 You're a professor at, is it York University? 18 00:01:27,608 --> 00:01:31,487 Osgood Hall Law School, which is part of York University in Toronto. 19 00:01:31,619 --> 00:01:34,166 Okay, and you teach legal innovation there. 20 00:01:34,166 --> 00:01:40,955 Why don't you tell us a little bit about what you are, I'm sorry, what you do and where you do it. 21 00:01:40,955 --> 00:01:41,955 Yeah, for sure. 22 00:01:41,955 --> 00:01:48,477 So yeah, I'm the National Director of AI Innovation and Knowledge at Dowlings where I just started last November. 23 00:01:48,477 --> 00:01:50,588 So still getting the lay of the land there. 24 00:01:50,588 --> 00:01:52,878 Lots of interesting stuff on the go. 25 00:01:53,339 --> 00:01:57,170 Adjunct professor at Osgoode where I teach a couple courses. 26 00:01:57,170 --> 00:02:04,142 One is focused on sort of design thinking methodology and innovation management generally, where we look at 27 00:02:04,142 --> 00:02:09,496 especially other industries, how innovation and disruptive technology have changed those industries. 28 00:02:09,496 --> 00:02:14,829 And then we bring applications to the practice of law and the business of law. 29 00:02:15,570 --> 00:02:21,013 The other course is just on pure legal tech and how lawyers can use that. 30 00:02:21,374 --> 00:02:26,287 And as you mentioned, I started at Norton Rose Fulbright. 31 00:02:26,287 --> 00:02:32,341 I guess I moved into the whole area of legal innovation almost by accident. 32 00:02:33,024 --> 00:02:42,058 I started as an entrepreneur when I was out of school, built a web development company which transitioned into a couple product companies that I sold. 33 00:02:42,058 --> 00:02:48,300 And sort of near the end of that, I was thinking about what do I want to do with the rest of my life? 34 00:02:48,620 --> 00:02:56,623 And in the greater Toronto area, there's a mode of transportation called the Go Train. 35 00:02:56,804 --> 00:03:02,626 And I remember looking out my window one day at all these people in suits. 36 00:03:02,850 --> 00:03:07,094 going to the GO train and getting on it and going to downtown Toronto. 37 00:03:07,094 --> 00:03:13,029 And I remember distinctly the thought, that's where the movers and the shakers are. 38 00:03:13,029 --> 00:03:20,865 Like that's where I need to go if I really want to interact with some interesting, influential people in society. 39 00:03:20,865 --> 00:03:22,866 So I thought, how do I get there? 40 00:03:23,007 --> 00:03:29,372 And the way that I thought is I should go to business school and figure out how I should have been running a business all along. 41 00:03:29,472 --> 00:03:31,628 And I ended up doing the joint. 42 00:03:31,628 --> 00:03:33,589 JD MBA program. 43 00:03:33,589 --> 00:03:41,701 So I went to law school at the same time, just thinking this may open some more more doors, no intention to practice law long term. 44 00:03:42,321 --> 00:03:47,433 But as you're part of law school, you get sort of caught up in the different recruitment cycles. 45 00:03:47,433 --> 00:03:50,604 And I thought, this might actually be interesting. 46 00:03:50,604 --> 00:03:54,765 And so I ended up going to Norton Rose and practicing there for a little while. 47 00:03:54,765 --> 00:03:58,946 And that's where I discovered this area of legal technology. 48 00:03:59,046 --> 00:04:00,507 and legal innovation. 49 00:04:00,507 --> 00:04:11,723 And so we sort of built out an innovation team from zero to around 11 people there where we did a lot of interesting things that are client facing and internal and all that kind 50 00:04:11,723 --> 00:04:12,694 of stuff. 51 00:04:13,454 --> 00:04:25,371 And so that's sort of the journey that I've taken from more of the pure tech entrepreneur perspective through to working in a large, very corporate structured environment for the 52 00:04:25,371 --> 00:04:27,042 last several years. 53 00:04:27,685 --> 00:04:28,125 Interesting. 54 00:04:28,125 --> 00:04:28,286 Yeah. 55 00:04:28,286 --> 00:04:31,448 A couple of things come to mind when you're describing your background. 56 00:04:31,448 --> 00:04:38,844 First, it's refreshing to hear that somebody like you is in an academic context teaching. 57 00:04:38,864 --> 00:04:44,028 I have a undergraduate degree in mathematics from UNC Chapel Hill. 58 00:04:44,029 --> 00:04:55,178 And then I started, actually started a business while I was an undergrad, a collection agency, which that's a story for another time, but it was a very interesting journey. 59 00:04:55,178 --> 00:04:57,103 And then I, 60 00:04:57,103 --> 00:05:01,325 went to work for Microsoft and then eventually bank of America. 61 00:05:01,325 --> 00:05:04,746 And while I was at bank of America, I got my MBA cause they paid for it. 62 00:05:04,746 --> 00:05:11,529 And there was a, I went to UNC Charlotte for my MBA, which is a part of the UNC system. 63 00:05:11,529 --> 00:05:15,951 But, they have a really heavy finance focus. 64 00:05:15,951 --> 00:05:20,373 Charlotte is actually the second largest banking town in the United States behind New York. 65 00:05:20,373 --> 00:05:24,435 Last time I checked, it was just ahead of San Francisco, but, um, 66 00:05:24,921 --> 00:05:27,712 My experience in the NBA was not a great one. 67 00:05:27,843 --> 00:05:33,176 the NBA program, I yeah, I found, well, a couple of things contributed to that. 68 00:05:33,176 --> 00:05:40,440 First, I had been an entrepreneur, so I had already, uh, you know, built a business from nothing. 69 00:05:40,440 --> 00:05:52,667 And, um, I mean, we were fairly successful, not really financially because it's, it's a low margin business, but we had 1100 clients in 23 States and, um, had a really big 70 00:05:52,667 --> 00:05:53,649 footprint. 71 00:05:53,649 --> 00:05:57,100 but there's really low barriers to entry in the collection business. 72 00:05:57,100 --> 00:06:02,441 And, um, it just wasn't, you really had to scale in order to be profitable. 73 00:06:02,541 --> 00:06:08,573 So I came to the table with a lot more experience than your average MBA student. 74 00:06:08,573 --> 00:06:22,867 So I found the, the, at the staff there was largely academic, a few, you know, like career academics, you know, like they did some consulting on the side, but 75 00:06:23,077 --> 00:06:24,788 man, so much. 76 00:06:24,788 --> 00:06:27,490 remember sitting in class going, yeah, that's not going to work. 77 00:06:27,490 --> 00:06:29,071 That's not going to work. 78 00:06:30,192 --> 00:06:33,405 and I got my real MBA in the trenches, right? 79 00:06:33,405 --> 00:06:35,206 Like actually doing it. 80 00:06:35,206 --> 00:06:45,024 So I'm, it's refreshing to hear that they're bringing somebody like you in who's actually done it and not just been an academic first and then done some consulting on the side, 81 00:06:45,024 --> 00:06:46,184 written some papers. 82 00:06:46,184 --> 00:06:49,977 Um, so I, I, what's that experience been like? 83 00:06:50,252 --> 00:06:51,553 Yeah, that's a great question. 84 00:06:51,553 --> 00:07:01,952 So I teach on the law school side, but a lot of the MBA methodology is the stuff that I'm bringing into the law school side. 85 00:07:01,952 --> 00:07:04,484 So I'll sort of get into some of the differences. 86 00:07:04,484 --> 00:07:12,021 When I went through the JD MBA program, I found the JD to be definitely more on the academic side. 87 00:07:12,021 --> 00:07:17,725 There's a very strict grading curve, highly competitive environment. 88 00:07:18,594 --> 00:07:22,046 but obviously great courses, great learnings and that kind of thing. 89 00:07:22,046 --> 00:07:25,618 On the MBA side, I actually found it to be a ton of fun. 90 00:07:25,718 --> 00:07:33,562 Lots of group projects, lots of creative exercises and interactive, a lot more hands-on. 91 00:07:33,734 --> 00:07:39,566 Many consulting projects are part of the curriculum of many of the courses there. 92 00:07:39,566 --> 00:07:44,109 So I sort of came from a different entrepreneurial background than you though, I think. 93 00:07:44,109 --> 00:07:46,170 I was sort of the lone 94 00:07:46,294 --> 00:07:54,139 entrepreneur in my basement, working with external consultants, not, you know, a team of 1100 people and that kind of thing. 95 00:07:54,139 --> 00:08:00,544 was definitely, I really wanted to go into that experience to start making connections. 96 00:08:00,544 --> 00:08:02,905 And I really liked that as a person. 97 00:08:02,905 --> 00:08:07,728 And I don't think it was really a good fit for me to be a lone tech entrepreneur anyway. 98 00:08:07,849 --> 00:08:11,801 And so for me, I absolutely loved all the creative stuff. 99 00:08:11,801 --> 00:08:13,353 I loved the group projects. 100 00:08:13,353 --> 00:08:15,854 And it's a lot of that, that I'm actually bringing 101 00:08:15,854 --> 00:08:23,694 into the law school side, where as we're talking about innovation, innovation is a very iterative process. 102 00:08:23,694 --> 00:08:32,134 You can't just have an idea and then start building it out and this is what is, you know, here's going to be my process and this will be my sales funnel and all that kind of stuff. 103 00:08:32,134 --> 00:08:43,168 You've got to get out there and get real feedback from people, which is completely different than the way many law school courses operate, which are a lot more. 104 00:08:43,168 --> 00:08:44,318 on the academic side. 105 00:08:44,318 --> 00:08:53,551 So going to a class and saying, and by the way, as part of your group project, I want you to pick up the phone and talk to people and get feedback on your idea. 106 00:08:53,551 --> 00:09:01,623 And you're probably going to be wrong a few times in terms of what you think is going to help a particular area of legal practice. 107 00:09:01,623 --> 00:09:04,154 But that's part of the innovation process. 108 00:09:04,154 --> 00:09:05,434 It's trial and error. 109 00:09:05,434 --> 00:09:07,234 It's people oriented. 110 00:09:07,234 --> 00:09:10,125 It's learning as you go kind of thing. 111 00:09:10,185 --> 00:09:13,092 And that's sort of in my experience for the 112 00:09:13,092 --> 00:09:15,999 several years in the teaching context. 113 00:09:16,391 --> 00:09:21,092 Well, it's good you decided not to be a tech entrepreneur because you've got great hair. 114 00:09:21,092 --> 00:09:22,573 I used to have great hair too. 115 00:09:22,573 --> 00:09:24,193 And now look at me. 116 00:09:24,193 --> 00:09:28,054 That's what happens when you start a tech business. 117 00:09:28,054 --> 00:09:30,215 You lose it, you go gray. 118 00:09:31,275 --> 00:09:37,927 But you know, it's funny, know, legal innovation to me and to many others, it's almost an oxymoron. 119 00:09:37,927 --> 00:09:41,418 It's like jumbo shrimp or military intelligence. 120 00:09:42,158 --> 00:09:44,931 There's a ton of innovation theater. 121 00:09:44,931 --> 00:09:51,283 that happens in legal and Mark Cohen wrote an awesome article on this subject. 122 00:09:51,283 --> 00:09:56,358 I don't know about two months ago and just how innovation is. 123 00:09:56,358 --> 00:10:02,601 It's not a project or an initiative or a department or a role on your org chart. 124 00:10:02,601 --> 00:10:11,986 It is a something that has to be adopted foundationally within the culture of the organization that's trying to innovate and law firms haven't done that. 125 00:10:12,773 --> 00:10:15,144 You know, many have dabbled. 126 00:10:15,485 --> 00:10:30,403 Even the, even the really innovation, innovative firms still have the structural limitations and impediments to innovation that are ultimately get in the way of really 127 00:10:30,403 --> 00:10:32,054 broad, innovative thinking. 128 00:10:32,054 --> 00:10:37,617 Like, and it's not to say, I'm definitely not saying there's not innovative work being done and legal. 129 00:10:37,617 --> 00:10:39,248 is for sure. 130 00:10:39,248 --> 00:10:41,467 And I see it and it's awesome. 131 00:10:41,467 --> 00:10:46,801 But when you talk about foundational, very little. 132 00:10:46,801 --> 00:10:52,736 I'm hoping that some of these barriers will... 133 00:10:52,736 --> 00:10:58,361 So the barriers that I have seen are, first of all, law firms don't like to spend on technology. 134 00:10:58,361 --> 00:10:59,762 They're laggards. 135 00:11:00,383 --> 00:11:07,008 They spend a minuscule percentage of revenue relative to other industries. 136 00:11:07,008 --> 00:11:09,120 Like this is an old number. 137 00:11:09,120 --> 00:11:10,531 I'm sure there are more... 138 00:11:10,585 --> 00:11:17,480 accurate numbers now, something like low single digit percentages of revenue go to tech spend. 139 00:11:17,480 --> 00:11:22,573 If you look at some industry like financial services, it's in the low to mid teens, right? 140 00:11:22,573 --> 00:11:28,307 So vastly different willingness to spend on technology. 141 00:11:28,768 --> 00:11:35,952 Leadership rises through the ranks by being the best at lawyering. 142 00:11:36,573 --> 00:11:39,775 You can't have professional management come in 143 00:11:39,929 --> 00:11:51,349 In law firms, they can't be owners, which is how you get people, uh, you know, through stock incentives and things other than just, know, their, their biweekly paycheck. 144 00:11:51,349 --> 00:12:01,248 Um, the economic model has been so successful for law firms because there really is much more supply than, than demand available. 145 00:12:01,248 --> 00:12:07,419 And the, the clients law firm clients have not demanded fundamental change. 146 00:12:07,419 --> 00:12:16,767 You know, they say, I want, you know, AFAs and you know, I want you to use AI to be more efficient, but they'll also say in the same sentence, yeah, I want to, I want to pay you a 147 00:12:16,767 --> 00:12:20,460 flat fee, but if it takes you less time, I want you to bill me hourly. 148 00:12:20,460 --> 00:12:23,312 And it's like, that's not, that doesn't, that doesn't work. 149 00:12:23,312 --> 00:12:24,574 So I don't know. 150 00:12:24,574 --> 00:12:30,218 What is your take on real innovation work happening in the legal space? 151 00:12:30,218 --> 00:12:31,499 Like foundational. 152 00:12:31,564 --> 00:12:43,861 Yeah, so I think there are some structural things you've mentioned are very, very important to the way in which innovation materializes within the industry. 153 00:12:43,861 --> 00:12:48,483 So for example, the ownership of law firms and all that, that kind of thing. 154 00:12:49,104 --> 00:12:51,895 I think that at some point that's going to change. 155 00:12:51,895 --> 00:12:58,989 We've certainly seen some differences in different jurisdictions around that kind of ownership model. 156 00:12:59,289 --> 00:13:01,130 And that is going to 157 00:13:01,388 --> 00:13:07,880 make a big effect on the way that innovation is handled, the way that we approach it. 158 00:13:08,140 --> 00:13:22,364 But given the structural constraints that we have now, I think the other thing to consider is that the legal industry is very much an apprenticeship industry versus other ones where 159 00:13:22,364 --> 00:13:29,528 there's a significant amount of spend in terms of the time that it takes to train juniors because they don't 160 00:13:29,528 --> 00:13:34,100 come out of law school knowing how to actually do the job itself. 161 00:13:34,100 --> 00:13:45,084 And so as we think about that sort of training process, there's great opportunity to insert innovation within that whole process where there's actually quite a big spend 162 00:13:45,084 --> 00:13:46,925 that's happening from law firms. 163 00:13:46,925 --> 00:13:57,009 So how does AI, how does innovation relate to that whole process where we're already spending in great amounts of time and resources? 164 00:13:57,229 --> 00:13:58,894 The other thing to think about 165 00:13:58,894 --> 00:14:11,874 is that although law school, or sorry, the legal industry is a trust industry, similar to financial institutions in that you don't really want your law firm going too far off the 166 00:14:11,874 --> 00:14:16,094 deep end and trying a whole bunch of risky things that maybe it'll work, maybe it won't. 167 00:14:16,094 --> 00:14:19,634 That's not really what clients want from a law firm. 168 00:14:19,914 --> 00:14:28,374 However, one of the advantages we have, and that allows, I think, the legal industry to move fast, even if we're not on the forefront, 169 00:14:28,374 --> 00:14:34,476 is that other industries are testing and trying a lot of the innovative technology. 170 00:14:34,476 --> 00:14:42,438 And the legal industry often is, you know, certainly not first to the party, but later on in that order. 171 00:14:42,438 --> 00:14:47,359 a lot of the risk in new technology has already been removed. 172 00:14:47,359 --> 00:14:52,871 And so that allows us to move a little bit faster while reducing that risk. 173 00:14:52,871 --> 00:14:57,858 So I think looking to other industries is really important in the legal industry because 174 00:14:57,858 --> 00:15:06,206 those are the ones that are taking those risks, failing in some cases, and then sort of testing what are the tools we can actually rely on. 175 00:15:06,587 --> 00:15:07,288 Yeah. 176 00:15:07,288 --> 00:15:16,074 Well, that's a good segue into what we were going to talk about in terms of AI and innovation and transformation. 177 00:15:17,356 --> 00:15:25,082 How do you see the role of AI in transforming knowledge management and library research functions? 178 00:15:25,082 --> 00:15:26,893 What's your perspective on that? 179 00:15:27,458 --> 00:15:40,888 Yeah, so I think there's two stages and this is not groundbreaking information here, but it's worth saying that the precursor to AI is really standardization and looking at what 180 00:15:40,888 --> 00:15:54,397 are the repeatable processes that we have, where is the consistent data that we can gather from those processes, where can we templatize things, these kinds of low tech activities 181 00:15:54,397 --> 00:15:55,916 are really the ones that are 182 00:15:55,916 --> 00:16:00,670 going to allow you to accelerate quite a bit when we get into AI. 183 00:16:00,670 --> 00:16:12,389 You you hear from either practitioners or just people in the market saying things like, let's just throw 300 documents into there and then say, now I want you to draft me 184 00:16:12,389 --> 00:16:16,702 something similar to that and make these changes. 185 00:16:17,263 --> 00:16:20,605 AI is just right now, it's not there yet, right? 186 00:16:20,605 --> 00:16:21,746 It's too... 187 00:16:22,104 --> 00:16:24,426 there's too much variance in terms of what you're going to get. 188 00:16:24,426 --> 00:16:30,219 It's not at the caliber of what someone would expect, especially of a large law firm. 189 00:16:30,480 --> 00:16:43,539 But if we take that extra step, if we do the spade work of really standardizing and looking at the repeatable processes and using knowledge management and research services 190 00:16:43,539 --> 00:16:50,508 and library services to get that standardized content, that is what's going to pour gasoline on 191 00:16:50,508 --> 00:16:56,166 what AI can do right now and we'll get a lot of efficiencies through that process. 192 00:16:56,668 --> 00:17:06,592 really the pathway to great disruption from AI is some of this low tech standardization work that needs to happen. 193 00:17:07,917 --> 00:17:11,489 IA before AI as they like to say, right? 194 00:17:12,270 --> 00:17:18,844 So yeah, at the KMNI conference last year, it's their second year. 195 00:17:18,844 --> 00:17:20,075 It's a great event. 196 00:17:20,075 --> 00:17:23,187 I learned so much from just sitting and listening. 197 00:17:23,187 --> 00:17:35,086 We sponsor, but I'm also an attendee and Sally Gonzalez from Fireman & Co did a, I think it might've been a keynote, one of the, maybe the first day keynote. 198 00:17:35,086 --> 00:17:37,687 And she outlined a framework. 199 00:17:37,815 --> 00:17:38,816 knowledge management. 200 00:17:38,816 --> 00:17:40,016 It was really good. 201 00:17:40,016 --> 00:17:42,817 Sally's been around since the beginning. 202 00:17:43,475 --> 00:17:50,435 I thought to myself, where does AI fit into that picture? 203 00:17:50,435 --> 00:17:54,552 I have some ideas on that, but I'm far removed. 204 00:17:54,552 --> 00:17:55,643 You're in the thick of it. 205 00:17:55,643 --> 00:18:04,707 How do you see law firms aligning like KM frameworks and the outcomes they want to achieve with AI? 206 00:18:05,966 --> 00:18:11,686 Yeah, so one treasure trove of opportunity is the document management system. 207 00:18:11,686 --> 00:18:22,345 And I've already described one way we can approach that is by looking at where's the activity in terms of the documents that are being generated, the documents that we're 208 00:18:22,345 --> 00:18:28,506 working on for clients, and looking at how can we standardize and templatize those things. 209 00:18:28,986 --> 00:18:34,444 Gowing really is a leader in that area of creating precedent material to really speed up 210 00:18:34,444 --> 00:18:35,174 efficiency. 211 00:18:35,174 --> 00:18:39,987 And so I'm really lucky coming into a firm like this, where we've got this huge head start. 212 00:18:39,987 --> 00:18:43,679 And I think AI is going to just accelerate that like crazy. 213 00:18:43,919 --> 00:18:55,275 The second area, though, is really around the data points that we can gather from the document management system from the structured information we have in our financial 214 00:18:55,275 --> 00:18:57,286 systems and that kind of thing. 215 00:18:57,306 --> 00:19:04,396 And it takes a little bit of curation and coordination in order to make sure that the data that 216 00:19:04,396 --> 00:19:10,438 these AI systems are going to sit on are cleaned up and are telling the right story and that kind of thing. 217 00:19:10,438 --> 00:19:15,166 I think knowledge management has a role to play in that curation process for sure. 218 00:19:16,027 --> 00:19:16,467 Yeah. 219 00:19:16,467 --> 00:19:18,848 Uh, in, in my experience. 220 00:19:18,848 --> 00:19:29,331 so between info dash and predecessor company Acrowire, we've worked with this number is bigger now, but a couple of years ago, it was like 110 AM law firms. 221 00:19:29,331 --> 00:19:35,733 So I've had a kind of a good cross-sectional view and the law firm DMS is typically a mess. 222 00:19:35,953 --> 00:19:43,685 And you know, there's all sorts of naming conventions, all sorts of different types of artifacts in it. 223 00:19:43,685 --> 00:19:46,237 There's multiple drafts. 224 00:19:46,237 --> 00:19:50,939 There's documents that were never used. 225 00:19:51,460 --> 00:20:01,105 It really feels like the curation, there needs to be some curation before we can leverage a DM in mass. 226 00:20:01,185 --> 00:20:04,140 But how do you, yeah. 227 00:20:04,140 --> 00:20:07,903 I think AI actually helps in that process as well. 228 00:20:08,004 --> 00:20:20,203 As we can deploy large language models on the document management system, which is something that every customer of a big DM is asking for either implicitly or explicitly to 229 00:20:20,203 --> 00:20:27,539 the vendors, as we can unleash AI broad scale across the DM and we know exactly what we're looking for. 230 00:20:27,539 --> 00:20:30,830 We know exactly how we want to tag certain types of documents. 231 00:20:30,830 --> 00:20:36,894 We know exactly the structured information that we want associated, like deal points and that kind of thing. 232 00:20:37,334 --> 00:20:40,957 Quite frankly, large language models can do that work. 233 00:20:40,957 --> 00:20:44,179 That works not that hard for the large language models to do. 234 00:20:44,179 --> 00:20:57,458 We just need a vendor to make those connections for us, to do that spade work of connecting the capabilities of large language models to this large scale of millions and 235 00:20:57,458 --> 00:21:00,930 millions of documents that can be under the direction 236 00:21:01,010 --> 00:21:06,738 of knowledge management professionals who will direct the AI to say, is what we're looking for. 237 00:21:06,738 --> 00:21:09,001 This is the structured information I need. 238 00:21:09,001 --> 00:21:16,200 And it's going to help that whole data cleaning and cleansing process quite a bit by leveraging AI on the front end. 239 00:21:16,683 --> 00:21:20,786 I'm of the opinion that it has to be the DM vendors that do that. 240 00:21:20,786 --> 00:21:40,650 And the reason is because of the scale of, and if you're in a cloud platform like I manage cloud or NetDocs, you can't interrogate a corpus of that size efficiently in any way. 241 00:21:40,650 --> 00:21:43,022 You have to be on top of it, right? 242 00:21:43,022 --> 00:21:45,113 Like infrastructure wise. 243 00:21:45,113 --> 00:21:53,227 It has to be very, very close, not over a WAN connection where the ingress and egress on that would be just a deal breaker. 244 00:21:53,227 --> 00:21:53,890 It'd be too slow. 245 00:21:53,890 --> 00:21:55,256 It'd be too expensive. 246 00:21:55,256 --> 00:21:57,298 It'd be very, very expensive. 247 00:21:57,298 --> 00:22:02,714 partially I understand why this hasn't been rolled out now, even though it's technically possible. 248 00:22:02,714 --> 00:22:08,630 We're waiting for prices to come down and more efficient reg pipelines in order to do it. 249 00:22:08,630 --> 00:22:18,720 But it's something I think everybody really needs and can leverage the document management system in a lot better ways once it happens. 250 00:22:18,961 --> 00:22:19,221 Yeah. 251 00:22:19,221 --> 00:22:20,282 And they're working on that. 252 00:22:20,282 --> 00:22:29,541 had Dan Hawk, the chief product officer from NetDocs on the podcast a couple months ago, and he described some ways that they're working on this. 253 00:22:29,541 --> 00:22:30,841 So they are. 254 00:22:31,663 --> 00:22:38,589 It's just, we're not there yet and it's still in flight. 255 00:22:38,589 --> 00:22:45,797 tell me about like, how should firms go about assessing the incremental value of 256 00:22:45,797 --> 00:22:49,008 these legal specific AI tools. 257 00:22:49,008 --> 00:22:49,908 There's so many. 258 00:22:49,908 --> 00:22:58,511 Like if I was in the CKO seat or the CINO seat, I would really be scratching my head how to tackle. 259 00:22:58,511 --> 00:23:02,112 There's so much out there and it's increasing every day. 260 00:23:02,112 --> 00:23:10,605 Like how should firms be thinking about finding the right solution for them and measuring incremental value? 261 00:23:11,995 --> 00:23:16,878 Well, I think first of all, firms need to actually ask that question. 262 00:23:17,378 --> 00:23:27,465 You know, just to sort of rewind a little bit, if you look at sort of the progression of the legal AI tools, which I think are wonderful and are very important, and I don't want 263 00:23:27,465 --> 00:23:29,546 to disparage them at all. 264 00:23:29,567 --> 00:23:36,571 I just really want to, I ask a lot of questions as a person, and so I think law firms should be asking a lot of questions as well. 265 00:23:36,571 --> 00:23:40,748 But if you look at the history, you know, we sort of went from zero 266 00:23:40,748 --> 00:23:44,341 to all of a sudden now we've got these legal AI tools, right? 267 00:23:44,341 --> 00:23:48,824 Like it happened really fast since the launch of the large language models, right? 268 00:23:48,824 --> 00:23:57,912 So a lot of people didn't have time to acclimate themselves in their personal life to making use of these large language models on a day-to-day basis. 269 00:23:57,912 --> 00:24:06,390 Early adopters like you and me were probably on chat, GPT and Claude and other tools right away, constantly using them, looking at the opportunities. 270 00:24:06,390 --> 00:24:10,871 in our own either personal lives or in a business and that kind of thing. 271 00:24:10,871 --> 00:24:19,303 And so then when we look at the legal large language models, model tools, you know, we're asking that question. 272 00:24:19,303 --> 00:24:27,546 So how is this incrementally better than what I could do in, you know, cloud or chat GPT or whatever, right? 273 00:24:27,546 --> 00:24:29,136 Like we're naturally asking that question. 274 00:24:29,136 --> 00:24:31,937 I think that law firms need to ask that question. 275 00:24:31,937 --> 00:24:36,268 And this is not to say that we would say, we don't need 276 00:24:36,418 --> 00:24:39,761 legal AI tools, I don't think that would be the conclusion. 277 00:24:39,761 --> 00:24:54,663 But it might be that there are certain use cases where we really do want to deploy a tool that has either training or they've nailed down the prompts under the hood to deal with 278 00:24:54,663 --> 00:24:56,314 legal use cases. 279 00:24:56,335 --> 00:24:59,878 you don't want to have to train lawyers on how to prompt effectively. 280 00:24:59,878 --> 00:25:03,761 You want it to be as natural as possible to the way they actually practice law. 281 00:25:03,761 --> 00:25:05,240 And the legal tools 282 00:25:05,240 --> 00:25:08,271 they really do help that in a lot of use cases. 283 00:25:08,451 --> 00:25:18,534 But then there are maybe other use cases where you might say, I don't know if there's an incremental value of the legal AI tool and maybe it would be better to have a secure 284 00:25:18,534 --> 00:25:25,055 version of the base model that is released within the firm for certain use cases. 285 00:25:25,356 --> 00:25:35,008 So I think these are important questions that need to be asked and every firm may come up with a different conclusion to that question. 286 00:25:35,205 --> 00:25:35,645 Yeah. 287 00:25:35,645 --> 00:25:41,100 Your point about experimenting on a personal basis is so incredibly important. 288 00:25:41,100 --> 00:25:50,938 Ethan Malik has a 10 hour challenge where he, you he challenges people to spend 10 hours performing tasks that they already do, but actually, you know, block time on your 289 00:25:50,938 --> 00:25:57,453 calendar, an hour every day for 10 days and run tasks through. 290 00:25:57,453 --> 00:26:01,677 Like a good example is how I prepare for this podcast. 291 00:26:01,677 --> 00:26:05,209 So I have a 30 minute planning call like you and I did. 292 00:26:05,443 --> 00:26:07,724 I download the transcript. 293 00:26:07,724 --> 00:26:11,416 have a custom Claude project with custom instructions. 294 00:26:11,416 --> 00:26:22,450 I've uploaded all sorts of KM specific material, like the skill last two years of the skills survey and like that outlines the priority because my audience is KM. 295 00:26:22,450 --> 00:26:28,613 want, I want my agenda, my agenda is to reflect the interests of the KM and I audience. 296 00:26:28,613 --> 00:26:33,703 Um, I uploaded examples of agendas that I wrote by hand. 297 00:26:33,703 --> 00:26:39,085 There's a couple of chapters from some books on knowledge management that I uploaded. 298 00:26:39,085 --> 00:26:53,812 And then I have custom instructions that tell this custom project how to go about its work, you know, to leverage the training materials first, to use a business casual tone, 299 00:26:54,212 --> 00:26:57,394 to align the topics to the training material. 300 00:26:57,394 --> 00:27:00,595 And I have taken, you know, how I used to do this was 301 00:27:00,667 --> 00:27:03,589 I would record the call and then go back and listen and take note. 302 00:27:03,589 --> 00:27:05,230 was ridiculous. 303 00:27:05,230 --> 00:27:16,815 Um, but I've saved so much time, but all of that process, like now I know how to build, I know how to build custom GPTs and Claude projects and I've incorporated them into my 304 00:27:16,815 --> 00:27:18,097 workflow. 305 00:27:18,097 --> 00:27:23,700 People need to understand how these things, and that's just, it's just a fancy rag implementation is all it is. 306 00:27:23,700 --> 00:27:26,021 Um, but I, I hear you, man. 307 00:27:26,021 --> 00:27:29,463 People have to spend time with the tools. 308 00:27:29,943 --> 00:27:35,426 and allocate resources towards that to really understand how it's going to apply in a work context. 309 00:27:35,426 --> 00:27:41,851 Yeah, and I would say that this is especially true of decision makers around technical purchases. 310 00:27:41,851 --> 00:27:47,036 I don't think the average person really is going to spend that time experimenting. 311 00:27:47,036 --> 00:27:51,979 I'm similar to you in that I use chat GPT all the time. 312 00:27:52,040 --> 00:27:59,606 And now that they've integrated with Siri where I don't have to have the voice mode open all the time to ask the one-off questions, you can just do it. 313 00:27:59,606 --> 00:28:02,168 It's even more the case. 314 00:28:02,168 --> 00:28:04,906 Literally just yesterday, I... 315 00:28:04,906 --> 00:28:10,479 you know, one use case for me in catching the train is I just want to know when the next trains are coming. 316 00:28:10,479 --> 00:28:13,198 And don't want to go through, open an app and this and that. 317 00:28:13,198 --> 00:28:14,320 And so I was sitting on the train. 318 00:28:14,320 --> 00:28:18,002 was like, why don't I just use generative AI here? 319 00:28:18,002 --> 00:28:23,484 Because I don't want to write the code that's going to go through a JSON object and figure out the times and that kind of thing. 320 00:28:23,484 --> 00:28:31,928 So I just opened the webpage, stuck it into a shortcut, and then just created a prompt, which you can do with shortcuts on the iPhone now. 321 00:28:31,928 --> 00:28:40,905 that says, look at this webpage, here are the data points that I want you to pull from it, and then tell me the next two trains, whether it's express and that kind of thing. 322 00:28:40,905 --> 00:28:46,729 But there's just so much low hanging fruit now that people can actually do. 323 00:28:46,830 --> 00:28:58,178 And we're in a totally different world that we have not yet seen the effects of these massive productivity gains across many different industries and many different areas. 324 00:28:58,683 --> 00:29:09,786 Yeah, we had a, we had a sales call yesterday with a firm who had their head in a place where they were going to do a custom dev project for their internet, which we did that for 325 00:29:09,786 --> 00:29:11,127 when we were known as Acrowire. 326 00:29:11,127 --> 00:29:12,867 We did that for like 12, 13 years. 327 00:29:12,867 --> 00:29:17,409 Like I know all the opportunities and challenges associated with that. 328 00:29:17,409 --> 00:29:21,430 And we decided to go down the product route because it's a better path. 329 00:29:21,470 --> 00:29:24,371 So, um, but right before the call, 330 00:29:24,571 --> 00:29:34,468 you know, I thought to myself, man, if I could use a metaphor here to describe, because you know, one of their concerns was, well, we don't know if, actually this was a, this was 331 00:29:34,468 --> 00:29:35,129 a different call. 332 00:29:35,129 --> 00:29:42,254 They, they wanted to, do a pilot and they weren't sure if their data was going to work with our tool. 333 00:29:42,254 --> 00:29:48,539 And I wanted to articulate to them that, you know, our solution is built for law firms and only law firms. 334 00:29:48,539 --> 00:29:50,030 And we've been doing it 16 years. 335 00:29:50,030 --> 00:29:52,155 So I got on chat GPT and advanced 336 00:29:52,155 --> 00:29:54,717 voice mode like three, four minutes before the call. 337 00:29:54,717 --> 00:30:04,445 And I said, Hey, I have a customer who is unsure about, um, how well their data is going to work with our solution. 338 00:30:04,445 --> 00:30:13,553 And I want to articulate a metaphor to help, you know, paint the picture and man, it nailed it. 339 00:30:13,553 --> 00:30:20,849 So it gave me a couple of great examples, like a, CUS a key custom cut for a lock that works every time a bespoke suit. 340 00:30:20,849 --> 00:30:22,401 gave me a whole bunch to choose from. 341 00:30:22,401 --> 00:30:26,526 And I was able to use that in my conversation. 342 00:30:26,526 --> 00:30:30,792 And it took me three, I started the process three, four minutes before the call. 343 00:30:30,792 --> 00:30:33,692 It's incredible what you can do these days. 344 00:30:33,692 --> 00:30:43,553 that's a great example for legal use cases actually, because what you and I are doing when we're using these models, we kind of have an understanding of what they can do and what 345 00:30:43,553 --> 00:30:44,574 they can't. 346 00:30:44,574 --> 00:30:53,403 And so we're able to guide and direct the tool to do what it's actually good at and then piece that together with the next component, right? 347 00:30:53,403 --> 00:30:56,766 This is sort of what I call the process of disaggregation. 348 00:30:56,896 --> 00:30:59,617 And that's what has to happen with the large language models. 349 00:30:59,617 --> 00:31:01,968 Now they can answer certain things, but not other things. 350 00:31:01,968 --> 00:31:05,449 They can process data and pull certain things out of it. 351 00:31:06,070 --> 00:31:08,110 Lawyers don't really think like that. 352 00:31:08,131 --> 00:31:20,356 And if, you think about for myself, having practiced at the junior level, there are some partners who, when they give you instructions, it's sort of like a high level thing. 353 00:31:20,356 --> 00:31:26,604 And then you've kind of got to go and figure out what does this even mean and start knocking on doors and asking people. 354 00:31:26,604 --> 00:31:31,566 And then there there's the rare partner that says, you know, these are the steps you're going to take. 355 00:31:31,566 --> 00:31:39,719 And after you get the result of that, then I want you to come back to me and we'll discuss that and then I'll send you off to the next piece. 356 00:31:39,719 --> 00:31:40,020 Right. 357 00:31:40,020 --> 00:31:52,265 So they're already disaggregating the overall task in their head so that a student isn't or a junior is not running off on rabbit trails, developing a completed project that makes 358 00:31:52,265 --> 00:31:55,336 no bloody sense whatsoever because they don't know what they're doing. 359 00:31:55,336 --> 00:31:56,096 Right. 360 00:31:56,364 --> 00:31:58,095 That's a large language model. 361 00:31:58,095 --> 00:32:04,339 We have to disaggregate the overall task and say, let's start with the creative process, right? 362 00:32:04,339 --> 00:32:07,221 I just need a metaphor and here are some examples, right? 363 00:32:07,221 --> 00:32:11,765 Then it comes back with some things and you say, hey, I really like the third one there. 364 00:32:11,765 --> 00:32:13,266 Let's work with that. 365 00:32:13,266 --> 00:32:17,128 And now the next step is I want to do A, B and C, right? 366 00:32:17,128 --> 00:32:21,431 Like that process of disaggregation and working iteratively with the tools. 367 00:32:21,431 --> 00:32:25,173 That is what I think is really going to change in the practice of law. 368 00:32:25,394 --> 00:32:26,284 But 369 00:32:26,518 --> 00:32:31,976 There's a learning process in terms of how lawyers are going to have to do that in the future. 370 00:32:32,327 --> 00:32:33,008 Yeah. 371 00:32:33,008 --> 00:32:41,914 Well, how, how should, um, you know, listeners of the podcast are probably sick of me talking about this, but it's, it's true. 372 00:32:41,914 --> 00:32:54,423 And so many firms are going a different direction, which is, know, when you're looking at defining an overall AI strategy for the firm, um, my, my recommendation is starting small 373 00:32:54,423 --> 00:32:56,804 and starting in the business of law. 374 00:32:56,945 --> 00:33:00,167 And the reason I think that makes the most sense. 375 00:33:00,167 --> 00:33:09,047 is first of all, business of law functions are the ones that are going to support the practice when the, when the practice of law solutions get deployed. 376 00:33:09,047 --> 00:33:10,867 So they need to have an understanding. 377 00:33:10,867 --> 00:33:13,607 So starting there is good for that. 378 00:33:13,607 --> 00:33:25,247 You don't have to worry about typically client data getting, you can use the public models for most of this stuff, like marketing, for example, a lot of times HR, like writing job 379 00:33:25,247 --> 00:33:29,847 descriptions, um, cut, you know, brainstorming comp plans. 380 00:33:30,147 --> 00:33:41,775 Um, finance, you know, there's all sorts of business of business of law functions that could, you could start small with and expand, um, incrementally. 381 00:33:41,775 --> 00:33:48,560 And I see a lot of firms not starting small, you know, they're jumping, you know, head first into Harvey or other tools. 382 00:33:48,560 --> 00:33:58,507 And, uh, you know, I, until recently, I knew very little about Harvey, but, uh, I had not seen a lot, but now they're starting to, and it's a, it's a great platform. 383 00:33:58,629 --> 00:34:05,244 I'm not saying don't do that, but I'm saying start small wherever you start. 384 00:34:05,244 --> 00:34:07,876 And business of law is not a bad place. 385 00:34:07,876 --> 00:34:09,567 What's your take on that? 386 00:34:09,742 --> 00:34:12,482 Okay, so I think there's two components here. 387 00:34:12,482 --> 00:34:16,422 One is start small, which I am 100 % on board with. 388 00:34:16,422 --> 00:34:28,342 And I'll sort of describe how I approach AI from a strategic standpoint, especially since I'm involved in talking about disruptive innovation across other industries in my teaching 389 00:34:28,342 --> 00:34:29,462 capacity. 390 00:34:29,782 --> 00:34:33,022 The second is the business of law component. 391 00:34:33,022 --> 00:34:36,182 So that's one way to start small for sure. 392 00:34:36,682 --> 00:34:39,922 The difficulty there, and I think you're right. 393 00:34:40,408 --> 00:34:47,113 But the difficulty there is when clients ask, how are you using AI to make things more efficient? 394 00:34:47,113 --> 00:34:50,466 And we say, oh, we're writing job descriptions with AI. 395 00:34:50,466 --> 00:34:54,129 like, oh, OK, really not that interesting kind of thing. 396 00:34:54,129 --> 00:35:06,319 So I think there's a pressure from the client side for sure for law firms to do the sexier, more innovative kind of uses of AI in the actual practice of law. 397 00:35:06,359 --> 00:35:10,412 But let's chat about that sort of start small. 398 00:35:10,890 --> 00:35:12,311 methodology. 399 00:35:12,351 --> 00:35:23,709 And, you know, when you look at the history of innovation and where disruptive technologies have come onto the scene, and there there's been winners and losers, some 400 00:35:23,709 --> 00:35:27,802 companies approached it really well, some companies didn't approach it so well. 401 00:35:27,802 --> 00:35:39,170 The ones that navigate disruptive technology really well, are the ones that look for immediate use cases of the technology that are very realistic. 402 00:35:39,170 --> 00:35:45,753 with respect to what the technology can actually do right then in the short term. 403 00:35:46,554 --> 00:35:57,120 Rather than trying to boil the ocean and, sometimes you win, sometimes you lose with that approach, but they look for those incremental innovations with the new technology. 404 00:35:57,120 --> 00:36:04,644 So an example that I use sometimes is when inkjet printers came out. 405 00:36:04,725 --> 00:36:07,042 Companies that navigated that really well, 406 00:36:07,042 --> 00:36:16,535 were ones where although they had perhaps a stranglehold over the laser printer market, they looked at this disruptive technology that was not good enough for their core 407 00:36:16,535 --> 00:36:17,545 customers. 408 00:36:17,545 --> 00:36:27,148 And so they didn't go out and say, okay, let's start pushing inkjet printers on all of our core customers and disrupt ourselves and that kind of thing. 409 00:36:27,328 --> 00:36:35,202 They decided let's create a new product line here that can operate independently and we can experiment. 410 00:36:35,202 --> 00:36:44,769 with this new technology and we'll let it grow and see where it goes rather than sort of sitting back, sticking our heads in the sands, letting other smaller companies begin using 411 00:36:44,769 --> 00:36:50,512 the disruptive technology and perhaps displacing us at some point. 412 00:36:50,533 --> 00:36:52,554 So let's keep it internal. 413 00:36:52,554 --> 00:36:54,208 Let's use it for what it's good at. 414 00:36:54,208 --> 00:36:58,879 This is not our core customer, but let's find the customers where this could actually help. 415 00:36:58,879 --> 00:37:04,402 And I think what law firms need to do now is look at when we're talking about AI in the practice of law, 416 00:37:04,406 --> 00:37:08,869 Certainly the business of law, there's tons of opportunities and I'm 100 % on board. 417 00:37:08,869 --> 00:37:10,770 We need to be doing that. 418 00:37:10,810 --> 00:37:19,905 But as we're looking at the practice of law, what we don't wanna do is throw AI at complex legal tasks that it's actually not very good at because lawyers are going to be very 419 00:37:19,905 --> 00:37:26,940 frustrated with trying to use a tool for something that it should not be used for right now. 420 00:37:26,940 --> 00:37:30,988 We need to look at those low level tasks. 421 00:37:30,988 --> 00:37:35,220 that lawyers are involved in in the process of producing legal work. 422 00:37:35,220 --> 00:37:47,797 Things like document review and the creative process in drafting and where you've got a narrow universe of content that you are plugging into the LLM to ground it in that 423 00:37:47,797 --> 00:37:50,609 content, whether it's a contract or a document. 424 00:37:50,609 --> 00:37:54,211 And there are very specific things that you are asking about it. 425 00:37:54,211 --> 00:38:00,014 Those are the use cases that generative AI is quite good at right now. 426 00:38:00,242 --> 00:38:03,946 And it's not everything that lawyers do, but we need to be realistic about that. 427 00:38:03,946 --> 00:38:09,010 Let's focus it in on what the AI is actually good at right now. 428 00:38:09,627 --> 00:38:10,248 Yeah. 429 00:38:10,248 --> 00:38:21,163 You know, trial prep is another use case that I think AI is outstanding for, you know, in a litigation scenario, you know, witness preparation. 430 00:38:21,222 --> 00:38:22,653 Yeah, what am I missing? 431 00:38:22,653 --> 00:38:26,196 Are there any contradictions in what the witness previously said? 432 00:38:26,196 --> 00:38:27,276 Those kinds of things. 433 00:38:27,276 --> 00:38:29,915 It's really, really useful for those things. 434 00:38:29,915 --> 00:38:33,047 Yeah, yeah, that's kind of low hanging fruit on the practice side. 435 00:38:33,047 --> 00:38:42,682 So you know what's interesting about your comment and I agree with you and I understand like RFPs coming out asking firms how they're using these technologies. 436 00:38:43,623 --> 00:38:47,155 Starting small should have started a year ago, right? 437 00:38:47,155 --> 00:38:51,867 So that that and obviously we can't go back in time. 438 00:38:51,867 --> 00:38:57,540 I understand that, but it's OK to start small aggressively, right? 439 00:38:57,540 --> 00:38:58,390 So. 440 00:38:58,919 --> 00:39:05,023 start small in the areas where you have the highest risk reward equation balance, right? 441 00:39:05,023 --> 00:39:06,664 That's most favorable. 442 00:39:06,804 --> 00:39:14,169 And in the practice of law, the use cases that we've been kicking around here are great examples. 443 00:39:14,470 --> 00:39:28,643 But you know what's funny is like you see this disparity, this disconnect, like some firms, and I think this has died down over time, but some firms in their OCGs required 444 00:39:28,643 --> 00:39:33,115 law firms to agree not to use the generative AI on their matters. 445 00:39:33,115 --> 00:39:36,356 And then you have other firms that are saying, how are you using it? 446 00:39:36,356 --> 00:39:36,956 You know what I mean? 447 00:39:36,956 --> 00:39:38,627 Like you need to be more efficient. 448 00:39:38,627 --> 00:39:44,638 there's been, I think over time that's that it's kind of like cloud, right? 449 00:39:44,638 --> 00:39:58,775 Like cloud, you know, um, it's exactly like cloud, you know, it's it, only thing is the clients aren't demanding that firms use cloud, but it was clients that held firms back. 450 00:39:58,929 --> 00:39:59,839 for the most part. 451 00:39:59,839 --> 00:40:09,177 And just there's, I've talked to some folks who had, you know, really kind of a skewed picture of the value prop, but yeah, it's like we. 452 00:40:09,223 --> 00:40:19,432 like in organizations, right, you hear the headlines of whatever company has seen proprietary code from a competitor and that kind of thing through using generative AI. 453 00:40:19,432 --> 00:40:29,140 And then that filters into these freak out sessions where, you know, we sort of think about what, what do we need to do to eliminate this risk with people who don't understand 454 00:40:29,140 --> 00:40:31,722 what the technology actually is or what it does. 455 00:40:31,722 --> 00:40:37,016 And then that goes through multiple levels where now that's been institutionalized and these are 456 00:40:37,016 --> 00:40:42,091 various guidelines and that kind of thing, which are for the most part, we're seeing less and less of it. 457 00:40:42,252 --> 00:40:48,118 However, I think what's always important with a new technology is to step back. 458 00:40:48,118 --> 00:40:51,041 And this is what I did three years ago at my previous firm. 459 00:40:51,041 --> 00:40:56,807 The question I kept asking is in five years, are we going to care about this? 460 00:40:56,807 --> 00:41:00,406 Is anyone going to care about these questions in five years? 461 00:41:00,406 --> 00:41:04,588 And that's not to say that, okay, so let's plow ahead and do it anyway. 462 00:41:04,588 --> 00:41:05,419 Absolutely not. 463 00:41:05,419 --> 00:41:07,000 You respect what the client wants. 464 00:41:07,000 --> 00:41:08,951 That's first and foremost. 465 00:41:08,951 --> 00:41:21,417 But as we're thinking strategically, as you're building your AI strategy within the organization, you have to have a long term enough vision that you're not building things 466 00:41:21,438 --> 00:41:28,001 and moving forward with initiatives that are just not going to matter in the medium term. 467 00:41:28,222 --> 00:41:28,602 Right? 468 00:41:28,602 --> 00:41:29,402 So 469 00:41:29,420 --> 00:41:32,871 I think that's really important as we think about strategy. 470 00:41:32,871 --> 00:41:45,015 And we've seen examples where people have looked at what can an LLM do now and why don't we train this on our data and we'll build our own LLM and do this and that, right? 471 00:41:45,015 --> 00:41:52,417 And then the new version of the model comes out and it blows away what everybody has been working on for the past year internally, right? 472 00:41:52,417 --> 00:41:56,792 It's again, because of a myopic vision of 473 00:41:56,792 --> 00:41:58,443 where the technology's at right now. 474 00:41:58,443 --> 00:42:08,830 You've got to at least look out five years in advance with any billed project or any significant process and think about where will the technology be then? 475 00:42:08,830 --> 00:42:10,121 What will people care about? 476 00:42:10,121 --> 00:42:13,113 How will we be practicing law at that point? 477 00:42:13,113 --> 00:42:21,218 And then how do we build our capacities now in order to prepare us for what law is going to look like in five and 10 years? 478 00:42:21,873 --> 00:42:23,443 makes perfect sense. 479 00:42:23,584 --> 00:42:37,984 And, you know, I haven't had a chance to, I'm not willing to write the $200 a month check to kick the tires on, uh, three yet, but I'm hearing from multiple sources that it, it 480 00:42:37,984 --> 00:42:40,768 really is incredible what it can do. 481 00:42:40,768 --> 00:42:49,141 And if you listen to Sam Altman and many others, they, there's a, there's a differing definitions on, on AGI and what that means. 482 00:42:49,141 --> 00:42:51,291 Uh, I don't know that there's really, 483 00:42:51,417 --> 00:42:56,468 any agreement, any any semblance of agreement on what age AGI is. 484 00:42:56,548 --> 00:43:13,213 But, um, what I, what I have heard is AGI that are a definition that makes sense to me is AGI is when AI can perform the overwhelming majority of tasks better than any human across 485 00:43:13,213 --> 00:43:14,413 any discipline. 486 00:43:14,413 --> 00:43:14,753 Right. 487 00:43:14,753 --> 00:43:18,354 That's, that sounds like a reasonable definition of AGI to me. 488 00:43:18,354 --> 00:43:21,297 What will that mean when we get there, whether it's 489 00:43:21,297 --> 00:43:25,652 You listen to Sam Altman and it's going to be the end of this year or you listen to others. 490 00:43:26,835 --> 00:43:28,167 you know, it'll be three years from now. 491 00:43:28,167 --> 00:43:33,183 What is that going to mean in a legal context and the practice of law? 492 00:43:33,210 --> 00:43:34,671 yeah, that's that's a question. 493 00:43:34,671 --> 00:43:37,092 I don't think I'm qualified to answer that. 494 00:43:37,092 --> 00:43:39,373 That's that's the question that we're asking, though. 495 00:43:39,373 --> 00:43:46,216 Like, what is law going to look like in a world where AI can do everything that a lawyer can just as well? 496 00:43:46,216 --> 00:43:53,069 And I think there's a couple of things that we're going to see in that kind of a world, assuming that world materializes. 497 00:43:53,069 --> 00:44:01,708 You know, as you're thinking about strategic planning, you plan for the most basic probabilities of what could happen. 498 00:44:01,708 --> 00:44:07,581 And certainly there's a possible world where what you just described actually happens, right? 499 00:44:07,581 --> 00:44:13,144 AI can do law just as well, if not better than people can. 500 00:44:13,144 --> 00:44:16,826 And we have to plan for that possibility. 501 00:44:17,166 --> 00:44:22,049 And I think in that kind of a world, the value proposition of a lawyer is very different. 502 00:44:22,049 --> 00:44:31,574 You know, the types of people that we want to hire are not necessarily the folks that are going to sit behind a computer screen and do technical areas of the law. 503 00:44:32,283 --> 00:44:35,986 what we look for in terms of grades and that kind of thing. 504 00:44:35,986 --> 00:44:47,686 It may change in terms of what we're looking for in lawyers coming out of law school because there's going to be a lot more complex, multidisciplinary problems where lawyers 505 00:44:47,686 --> 00:44:58,085 function a lot more like consultants, using AI to do the technical stuff and then solving these business problems, these complex business problems for their clients. 506 00:44:58,149 --> 00:44:59,039 Yeah. 507 00:44:59,039 --> 00:45:02,012 Legal today is so work product focused, right? 508 00:45:02,012 --> 00:45:03,283 That's the deliverable. 509 00:45:03,283 --> 00:45:12,650 And the reality is that lawyers have so much more to offer than just a word or a PDF document. 510 00:45:12,650 --> 00:45:13,771 It's their knowledge. 511 00:45:13,771 --> 00:45:14,671 It's their context. 512 00:45:14,671 --> 00:45:15,632 It's their understanding. 513 00:45:15,632 --> 00:45:17,714 It's their ability to connect dots. 514 00:45:17,714 --> 00:45:25,563 It's their ability to project in the future, knowing like from a regulatory perspective, what's in queue and what's likely to come. 515 00:45:25,563 --> 00:45:29,635 to fruition with the new administration, for example, like that sort of stuff. 516 00:45:29,635 --> 00:45:30,917 AI can't do that. 517 00:45:30,917 --> 00:45:31,147 Right. 518 00:45:31,147 --> 00:45:40,835 Because there's reading of tea leaves that has to happen there and extrapolation, and it would be very difficult, um, to pull all those pieces together. 519 00:45:40,835 --> 00:45:41,596 So I agree with you. 520 00:45:41,596 --> 00:45:48,802 It's going to turn more into a consulting business than just a pure, you know, work product deliverable scenario. 521 00:45:48,802 --> 00:45:50,863 Like it happens a lot today. 522 00:45:51,246 --> 00:45:53,626 And I think there's a couple other things as well. 523 00:45:53,626 --> 00:45:57,906 So one is the role of a lawyer is definitely going to change. 524 00:45:57,906 --> 00:46:07,965 The other thing I think law firms, especially large firms, are going to be tasked with handling mass, absolutely astronomical amounts of data. 525 00:46:08,046 --> 00:46:16,146 If you think about what AI enables in our legal system, Our legal system, the sort of the pinnacle of it is the courts, right? 526 00:46:16,146 --> 00:46:19,926 That's sort of what drives everything where decisions are made. 527 00:46:20,342 --> 00:46:25,367 Right now there is an artificial barrier to entry into the whole legal system. 528 00:46:25,367 --> 00:46:32,634 It's quite expensive to hire a lawyer and I've got a dispute or whatever it is. 529 00:46:32,855 --> 00:46:39,842 Once AI can do all of that stuff as well as a person, there's no more barrier to entry anymore. 530 00:46:39,842 --> 00:46:44,066 Our legal system is going to be completely flooded with cases. 531 00:46:44,318 --> 00:46:53,132 large corporate clients are going to be flooded with these cases that are coming in and they're going to need law firms with the technical capabilities to be able to handle 532 00:46:53,132 --> 00:46:55,183 massive amounts of data. 533 00:46:55,183 --> 00:47:05,447 So not only is the legal system going to be flooded for that, but even on the enforcement side, enforcement bodies are going to be able to use AI to find infractions that right now 534 00:47:05,447 --> 00:47:10,949 would take months and months and months of combing through paperwork in order to find the infraction. 535 00:47:10,949 --> 00:47:12,706 They'll be able to find it instantly. 536 00:47:12,706 --> 00:47:14,176 the risk of clients is going up. 537 00:47:14,176 --> 00:47:18,451 So there's just going to be a ton of work for law firms to actually navigate. 538 00:47:18,451 --> 00:47:28,679 So you've got this multidisciplinary consulting happening with this increase in volume, like a huge increase in the volume of work. 539 00:47:28,679 --> 00:47:36,165 So I think there are going to be different ways that law firms as almost a consulting organization are going to be able to deal with that. 540 00:47:36,326 --> 00:47:42,094 One is sort of the personal consulting approach, but the other one might be 541 00:47:42,094 --> 00:47:54,931 productized offerings that, you know, clients can use instantly in order to help them navigate things that leverage the data that law firms are going to accumulate through this 542 00:47:54,931 --> 00:47:56,402 increase in legal work. 543 00:47:56,635 --> 00:47:57,535 Yeah. 544 00:47:57,556 --> 00:48:09,170 And I know we're almost out of time, but one thing I wanted to quickly ask you about, which you alluded to in your intro was lessons from other industries. 545 00:48:09,170 --> 00:48:16,077 Like what can we learn from how other industries have managed disruption and innovation effectively? 546 00:48:16,077 --> 00:48:17,879 What are your, what are your thoughts on that? 547 00:48:18,082 --> 00:48:20,514 Yeah, I touched on this a little bit previously. 548 00:48:20,514 --> 00:48:29,252 So I think the starting point is to be realistic about where the technology is now, but also very cognizant of where the technology is going. 549 00:48:29,252 --> 00:48:33,145 And we've gone, you know, both these places in our conversation, right? 550 00:48:33,145 --> 00:48:39,901 Like AI, it's not doing the high value stuff right now, but there's a possible world where it does do that. 551 00:48:39,901 --> 00:48:46,586 So then the question is strategically for an organization, how do I leverage what it can do now? 552 00:48:46,742 --> 00:48:51,424 and build a blueprint for where the technology is going. 553 00:48:51,424 --> 00:48:58,527 So an example that I think I and probably others use all the time is that of Netflix, right? 554 00:48:58,527 --> 00:49:03,969 When Netflix built its company, it was on version one of the internet. 555 00:49:03,969 --> 00:49:07,180 There was no streaming, broadband wasn't happening. 556 00:49:07,180 --> 00:49:14,413 Like the technical capabilities were not there for an on-demand video delivery service, right? 557 00:49:14,693 --> 00:49:16,482 But strategically, 558 00:49:16,482 --> 00:49:22,264 They built their company to win in that world that was very quickly coming. 559 00:49:22,365 --> 00:49:25,726 So what do you need to win in a streaming world? 560 00:49:25,726 --> 00:49:26,986 You need a customer base. 561 00:49:26,986 --> 00:49:29,117 You need algorithms for recommendation. 562 00:49:29,117 --> 00:49:36,781 You need to get people used to clicking on websites in order to navigate their video rentals rather than going into a store. 563 00:49:36,781 --> 00:49:38,621 You need all these different capacities. 564 00:49:38,621 --> 00:49:40,232 You need an inventory catalog. 565 00:49:40,232 --> 00:49:43,133 You need licensing agreements with different companies. 566 00:49:43,133 --> 00:49:45,474 All this stuff is necessary. 567 00:49:45,518 --> 00:49:47,159 for the streaming world. 568 00:49:47,419 --> 00:49:48,380 But what did they do? 569 00:49:48,380 --> 00:49:53,943 They built a mail order business because that's what this technology could support at the time. 570 00:49:53,943 --> 00:50:02,848 And it wasn't as good as brick and mortar in some ways, but they made use of what the technology could do in the present in order to win the future. 571 00:50:02,888 --> 00:50:08,831 And that's really why I think it's important that law firms look at what can AI do right now? 572 00:50:08,831 --> 00:50:13,934 And how can we use that as a blueprint in order to win a future? 573 00:50:13,954 --> 00:50:17,437 where AI can lawyer just as well as people can. 574 00:50:18,011 --> 00:50:18,361 Yeah. 575 00:50:18,361 --> 00:50:29,166 And you know, what's another interesting aspect of Netflix and what they were able to accomplish is they started streaming other people's content and then they, they saw the 576 00:50:29,166 --> 00:50:41,641 writing on the wall of, you know, Disney and NBC and all these content producers creating their own networks and they got ahead of that and started doing original content. 577 00:50:41,641 --> 00:50:46,853 So yeah, kudos to Netflix for, you know, seeing 578 00:50:46,853 --> 00:50:49,525 the future and preparing for it proactively. 579 00:50:49,525 --> 00:50:59,974 They didn't wait until these content providers turn the screws on the licensing agreements to make their business unviable. 580 00:50:59,974 --> 00:51:06,849 They started early and got ahead of it and have been wildly successful as a result of that. 581 00:51:07,190 --> 00:51:09,201 It's what every successful company has done. 582 00:51:09,201 --> 00:51:14,664 You look at Uber, they didn't build their company in order to create a gig economy. 583 00:51:14,664 --> 00:51:16,395 They happened to do that. 584 00:51:16,395 --> 00:51:22,819 They built their company because of self-driving vehicles that is the future we're moving into, right? 585 00:51:22,819 --> 00:51:24,730 That's where the profitability is, right? 586 00:51:24,730 --> 00:51:27,662 So you look to the future, where are we going? 587 00:51:27,662 --> 00:51:31,384 But what do I need to build in the present in order to win that future? 588 00:51:31,384 --> 00:51:35,266 Because I need a customer base in order to win. 589 00:51:35,266 --> 00:51:39,043 the self-driving fleet possible future, right? 590 00:51:39,043 --> 00:51:44,252 So what do law firms need to build right now to win the future of AI lawyering? 591 00:51:46,299 --> 00:51:48,860 Yeah, that's a great way to end it. 592 00:51:48,860 --> 00:52:04,564 Those are very powerful analogies and ways to think about how law firms need to start thinking foundationally different like Uber and Netflix because that level of 593 00:52:04,564 --> 00:52:11,916 transformation is coming and the firms that prepare for it will be successful. 594 00:52:11,916 --> 00:52:15,597 Those that don't will be the blockbusters of the world. 595 00:52:17,570 --> 00:52:18,687 Very true. 596 00:52:19,316 --> 00:52:22,137 well, it's been a great conversation. 597 00:52:22,137 --> 00:52:25,118 We've been, again, talking about this forever. 598 00:52:25,118 --> 00:52:36,613 You transitioned roles, so I wanted to give you some time to get your feet underneath you there, but we really look forward to working with you at Gowling, and I appreciate the 599 00:52:36,613 --> 00:52:37,623 time today. 600 00:52:37,634 --> 00:52:39,711 Yeah, thanks Ted, thanks for having me. 601 00:52:40,953 --> 00:52:41,689 Absolutely. 602 00:52:41,689 --> 00:52:42,091 All right. 603 00:52:42,091 --> 00:52:43,094 Have a good one. 604 00:52:43,310 --> 00:52:44,169 You too. 00:00:03,640 Al, good to see you this morning. 2 00:00:03,682 --> 00:00:05,055 Good to see you, Ted. 3 00:00:05,777 --> 00:00:09,950 So you and I talked about getting together on a podcast. 4 00:00:09,950 --> 00:00:19,237 think it was at evolve last year when we both won basically the same award, but you were on the law firm side. 5 00:00:19,237 --> 00:00:27,203 I was on the vendor side and I think we're still the reigning innovators of the year for a few more months. 6 00:00:29,926 --> 00:00:31,087 Exactly. 7 00:00:31,087 --> 00:00:32,399 So, um, 8 00:00:32,399 --> 00:00:33,360 Yeah. 9 00:00:33,360 --> 00:00:38,042 What's interesting is, you came from NRF, right? 10 00:00:38,042 --> 00:00:50,920 You were NRF Canada and we just made NRF US a client of InfoDash and now you're at Gowling, who's also a client of InfoDash, an early adopter and you guys are doing some 11 00:00:50,920 --> 00:00:51,971 really cool stuff. 12 00:00:51,971 --> 00:01:00,185 I hope you guys apply for, I don't know if you've had a chance to see the internet there, but y'all did an amazing job with the layout and design. 13 00:01:00,396 --> 00:01:04,443 Yeah, I just, actually had a demo this week, so really happy with it. 14 00:01:04,443 --> 00:01:06,504 Yeah, it looks really good. 15 00:01:07,765 --> 00:01:12,647 So yeah, before we jump into the agenda, let's get you introduced. 16 00:01:13,428 --> 00:01:22,192 You're now the National Director of AI, Innovation and Knowledge at Gowling. 17 00:01:22,433 --> 00:01:27,515 You're a professor at, is it York University? 18 00:01:27,608 --> 00:01:31,487 Osgood Hall Law School, which is part of York University in Toronto. 19 00:01:31,619 --> 00:01:34,166 Okay, and you teach legal innovation there. 20 00:01:34,166 --> 00:01:40,955 Why don't you tell us a little bit about what you are, I'm sorry, what you do and where you do it. 21 00:01:40,955 --> 00:01:41,955 Yeah, for sure. 22 00:01:41,955 --> 00:01:48,477 So yeah, I'm the National Director of AI Innovation and Knowledge at Dowlings where I just started last November. 23 00:01:48,477 --> 00:01:50,588 So still getting the lay of the land there. 24 00:01:50,588 --> 00:01:52,878 Lots of interesting stuff on the go. 25 00:01:53,339 --> 00:01:57,170 Adjunct professor at Osgoode where I teach a couple courses. 26 00:01:57,170 --> 00:02:04,142 One is focused on sort of design thinking methodology and innovation management generally, where we look at 27 00:02:04,142 --> 00:02:09,496 especially other industries, how innovation and disruptive technology have changed those industries. 28 00:02:09,496 --> 00:02:14,829 And then we bring applications to the practice of law and the business of law. 29 00:02:15,570 --> 00:02:21,013 The other course is just on pure legal tech and how lawyers can use that. 30 00:02:21,374 --> 00:02:26,287 And as you mentioned, I started at Norton Rose Fulbright. 31 00:02:26,287 --> 00:02:32,341 I guess I moved into the whole area of legal innovation almost by accident. 32 00:02:33,024 --> 00:02:42,058 I started as an entrepreneur when I was out of school, built a web development company which transitioned into a couple product companies that I sold. 33 00:02:42,058 --> 00:02:48,300 And sort of near the end of that, I was thinking about what do I want to do with the rest of my life? 34 00:02:48,620 --> 00:02:56,623 And in the greater Toronto area, there's a mode of transportation called the Go Train. 35 00:02:56,804 --> 00:03:02,626 And I remember looking out my window one day at all these people in suits. 36 00:03:02,850 --> 00:03:07,094 going to the GO train and getting on it and going to downtown Toronto. 37 00:03:07,094 --> 00:03:13,029 And I remember distinctly the thought, that's where the movers and the shakers are. 38 00:03:13,029 --> 00:03:20,865 Like that's where I need to go if I really want to interact with some interesting, influential people in society. 39 00:03:20,865 --> 00:03:22,866 So I thought, how do I get there? 40 00:03:23,007 --> 00:03:29,372 And the way that I thought is I should go to business school and figure out how I should have been running a business all along. 41 00:03:29,472 --> 00:03:31,628 And I ended up doing the joint. 42 00:03:31,628 --> 00:03:33,589 JD MBA program. 43 00:03:33,589 --> 00:03:41,701 So I went to law school at the same time, just thinking this may open some more more doors, no intention to practice law long term. 44 00:03:42,321 --> 00:03:47,433 But as you're part of law school, you get sort of caught up in the different recruitment cycles. 45 00:03:47,433 --> 00:03:50,604 And I thought, this might actually be interesting. 46 00:03:50,604 --> 00:03:54,765 And so I ended up going to Norton Rose and practicing there for a little while. 47 00:03:54,765 --> 00:03:58,946 And that's where I discovered this area of legal technology. 48 00:03:59,046 --> 00:04:00,507 and legal innovation. 49 00:04:00,507 --> 00:04:11,723 And so we sort of built out an innovation team from zero to around 11 people there where we did a lot of interesting things that are client facing and internal and all that kind 50 00:04:11,723 --> 00:04:12,694 of stuff. 51 00:04:13,454 --> 00:04:25,371 And so that's sort of the journey that I've taken from more of the pure tech entrepreneur perspective through to working in a large, very corporate structured environment for the 52 00:04:25,371 --> 00:04:27,042 last several years. 53 00:04:27,685 --> 00:04:28,125 Interesting. 54 00:04:28,125 --> 00:04:28,286 Yeah. 55 00:04:28,286 --> 00:04:31,448 A couple of things come to mind when you're describing your background. 56 00:04:31,448 --> 00:04:38,844 First, it's refreshing to hear that somebody like you is in an academic context teaching. 57 00:04:38,864 --> 00:04:44,028 I have a undergraduate degree in mathematics from UNC Chapel Hill. 58 00:04:44,029 --> 00:04:55,178 And then I started, actually started a business while I was an undergrad, a collection agency, which that's a story for another time, but it was a very interesting journey. 59 00:04:55,178 --> 00:04:57,103 And then I, 60 00:04:57,103 --> 00:05:01,325 went to work for Microsoft and then eventually bank of America. 61 00:05:01,325 --> 00:05:04,746 And while I was at bank of America, I got my MBA cause they paid for it. 62 00:05:04,746 --> 00:05:11,529 And there was a, I went to UNC Charlotte for my MBA, which is a part of the UNC system. 63 00:05:11,529 --> 00:05:15,951 But, they have a really heavy finance focus. 64 00:05:15,951 --> 00:05:20,373 Charlotte is actually the second largest banking town in the United States behind New York. 65 00:05:20,373 --> 00:05:24,435 Last time I checked, it was just ahead of San Francisco, but, um, 66 00:05:24,921 --> 00:05:27,712 My experience in the NBA was not a great one. 67 00:05:27,843 --> 00:05:33,176 the NBA program, I yeah, I found, well, a couple of things contributed to that. 68 00:05:33,176 --> 00:05:40,440 First, I had been an entrepreneur, so I had already, uh, you know, built a business from nothing. 69 00:05:40,440 --> 00:05:52,667 And, um, I mean, we were fairly successful, not really financially because it's, it's a low margin business, but we had 1100 clients in 23 States and, um, had a really big 70 00:05:52,667 --> 00:05:53,649 footprint. 71 00:05:53,649 --> 00:05:57,100 but there's really low barriers to entry in the collection business. 72 00:05:57,100 --> 00:06:02,441 And, um, it just wasn't, you really had to scale in order to be profitable. 73 00:06:02,541 --> 00:06:08,573 So I came to the table with a lot more experience than your average MBA student. 74 00:06:08,573 --> 00:06:22,867 So I found the, the, at the staff there was largely academic, a few, you know, like career academics, you know, like they did some consulting on the side, but 75 00:06:23,077 --> 00:06:24,788 man, so much. 76 00:06:24,788 --> 00:06:27,490 remember sitting in class going, yeah, that's not going to work. 77 00:06:27,490 --> 00:06:29,071 That's not going to work. 78 00:06:30,192 --> 00:06:33,405 and I got my real MBA in the trenches, right? 79 00:06:33,405 --> 00:06:35,206 Like actually doing it. 80 00:06:35,206 --> 00:06:45,024 So I'm, it's refreshing to hear that they're bringing somebody like you in who's actually done it and not just been an academic first and then done some consulting on the side, 81 00:06:45,024 --> 00:06:46,184 written some papers. 82 00:06:46,184 --> 00:06:49,977 Um, so I, I, what's that experience been like? 83 00:06:50,252 --> 00:06:51,553 Yeah, that's a great question. 84 00:06:51,553 --> 00:07:01,952 So I teach on the law school side, but a lot of the MBA methodology is the stuff that I'm bringing into the law school side. 85 00:07:01,952 --> 00:07:04,484 So I'll sort of get into some of the differences. 86 00:07:04,484 --> 00:07:12,021 When I went through the JD MBA program, I found the JD to be definitely more on the academic side. 87 00:07:12,021 --> 00:07:17,725 There's a very strict grading curve, highly competitive environment. 88 00:07:18,594 --> 00:07:22,046 but obviously great courses, great learnings and that kind of thing. 89 00:07:22,046 --> 00:07:25,618 On the MBA side, I actually found it to be a ton of fun. 90 00:07:25,718 --> 00:07:33,562 Lots of group projects, lots of creative exercises and interactive, a lot more hands-on. 91 00:07:33,734 --> 00:07:39,566 Many consulting projects are part of the curriculum of many of the courses there. 92 00:07:39,566 --> 00:07:44,109 So I sort of came from a different entrepreneurial background than you though, I think. 93 00:07:44,109 --> 00:07:46,170 I was sort of the lone 94 00:07:46,294 --> 00:07:54,139 entrepreneur in my basement, working with external consultants, not, you know, a team of 1100 people and that kind of thing. 95 00:07:54,139 --> 00:08:00,544 was definitely, I really wanted to go into that experience to start making connections. 96 00:08:00,544 --> 00:08:02,905 And I really liked that as a person. 97 00:08:02,905 --> 00:08:07,728 And I don't think it was really a good fit for me to be a lone tech entrepreneur anyway. 98 00:08:07,849 --> 00:08:11,801 And so for me, I absolutely loved all the creative stuff. 99 00:08:11,801 --> 00:08:13,353 I loved the group projects. 100 00:08:13,353 --> 00:08:15,854 And it's a lot of that, that I'm actually bringing 101 00:08:15,854 --> 00:08:23,694 into the law school side, where as we're talking about innovation, innovation is a very iterative process. 102 00:08:23,694 --> 00:08:32,134 You can't just have an idea and then start building it out and this is what is, you know, here's going to be my process and this will be my sales funnel and all that kind of stuff. 103 00:08:32,134 --> 00:08:43,168 You've got to get out there and get real feedback from people, which is completely different than the way many law school courses operate, which are a lot more. 104 00:08:43,168 --> 00:08:44,318 on the academic side. 105 00:08:44,318 --> 00:08:53,551 So going to a class and saying, and by the way, as part of your group project, I want you to pick up the phone and talk to people and get feedback on your idea. 106 00:08:53,551 --> 00:09:01,623 And you're probably going to be wrong a few times in terms of what you think is going to help a particular area of legal practice. 107 00:09:01,623 --> 00:09:04,154 But that's part of the innovation process. 108 00:09:04,154 --> 00:09:05,434 It's trial and error. 109 00:09:05,434 --> 00:09:07,234 It's people oriented. 110 00:09:07,234 --> 00:09:10,125 It's learning as you go kind of thing. 111 00:09:10,185 --> 00:09:13,092 And that's sort of in my experience for the 112 00:09:13,092 --> 00:09:15,999 several years in the teaching context. 113 00:09:16,391 --> 00:09:21,092 Well, it's good you decided not to be a tech entrepreneur because you've got great hair. 114 00:09:21,092 --> 00:09:22,573 I used to have great hair too. 115 00:09:22,573 --> 00:09:24,193 And now look at me. 116 00:09:24,193 --> 00:09:28,054 That's what happens when you start a tech business. 117 00:09:28,054 --> 00:09:30,215 You lose it, you go gray. 118 00:09:31,275 --> 00:09:37,927 But you know, it's funny, know, legal innovation to me and to many others, it's almost an oxymoron. 119 00:09:37,927 --> 00:09:41,418 It's like jumbo shrimp or military intelligence. 120 00:09:42,158 --> 00:09:44,931 There's a ton of innovation theater. 121 00:09:44,931 --> 00:09:51,283 that happens in legal and Mark Cohen wrote an awesome article on this subject. 122 00:09:51,283 --> 00:09:56,358 I don't know about two months ago and just how innovation is. 123 00:09:56,358 --> 00:10:02,601 It's not a project or an initiative or a department or a role on your org chart. 124 00:10:02,601 --> 00:10:11,986 It is a something that has to be adopted foundationally within the culture of the organization that's trying to innovate and law firms haven't done that. 125 00:10:12,773 --> 00:10:15,144 You know, many have dabbled. 126 00:10:15,485 --> 00:10:30,403 Even the, even the really innovation, innovative firms still have the structural limitations and impediments to innovation that are ultimately get in the way of really 127 00:10:30,403 --> 00:10:32,054 broad, innovative thinking. 128 00:10:32,054 --> 00:10:37,617 Like, and it's not to say, I'm definitely not saying there's not innovative work being done and legal. 129 00:10:37,617 --> 00:10:39,248 is for sure. 130 00:10:39,248 --> 00:10:41,467 And I see it and it's awesome. 131 00:10:41,467 --> 00:10:46,801 But when you talk about foundational, very little. 132 00:10:46,801 --> 00:10:52,736 I'm hoping that some of these barriers will... 133 00:10:52,736 --> 00:10:58,361 So the barriers that I have seen are, first of all, law firms don't like to spend on technology. 134 00:10:58,361 --> 00:10:59,762 They're laggards. 135 00:11:00,383 --> 00:11:07,008 They spend a minuscule percentage of revenue relative to other industries. 136 00:11:07,008 --> 00:11:09,120 Like this is an old number. 137 00:11:09,120 --> 00:11:10,531 I'm sure there are more... 138 00:11:10,585 --> 00:11:17,480 accurate numbers now, something like low single digit percentages of revenue go to tech spend. 139 00:11:17,480 --> 00:11:22,573 If you look at some industry like financial services, it's in the low to mid teens, right? 140 00:11:22,573 --> 00:11:28,307 So vastly different willingness to spend on technology. 141 00:11:28,768 --> 00:11:35,952 Leadership rises through the ranks by being the best at lawyering. 142 00:11:36,573 --> 00:11:39,775 You can't have professional management come in 143 00:11:39,929 --> 00:11:51,349 In law firms, they can't be owners, which is how you get people, uh, you know, through stock incentives and things other than just, know, their, their biweekly paycheck. 144 00:11:51,349 --> 00:12:01,248 Um, the economic model has been so successful for law firms because there really is much more supply than, than demand available. 145 00:12:01,248 --> 00:12:07,419 And the, the clients law firm clients have not demanded fundamental change. 146 00:12:07,419 --> 00:12:16,767 You know, they say, I want, you know, AFAs and you know, I want you to use AI to be more efficient, but they'll also say in the same sentence, yeah, I want to, I want to pay you a 147 00:12:16,767 --> 00:12:20,460 flat fee, but if it takes you less time, I want you to bill me hourly. 148 00:12:20,460 --> 00:12:23,312 And it's like, that's not, that doesn't, that doesn't work. 149 00:12:23,312 --> 00:12:24,574 So I don't know. 150 00:12:24,574 --> 00:12:30,218 What is your take on real innovation work happening in the legal space? 151 00:12:30,218 --> 00:12:31,499 Like foundational. 152 00:12:31,564 --> 00:12:43,861 Yeah, so I think there are some structural things you've mentioned are very, very important to the way in which innovation materializes within the industry. 153 00:12:43,861 --> 00:12:48,483 So for example, the ownership of law firms and all that, that kind of thing. 154 00:12:49,104 --> 00:12:51,895 I think that at some point that's going to change. 155 00:12:51,895 --> 00:12:58,989 We've certainly seen some differences in different jurisdictions around that kind of ownership model. 156 00:12:59,289 --> 00:13:01,130 And that is going to 157 00:13:01,388 --> 00:13:07,880 make a big effect on the way that innovation is handled, the way that we approach it. 158 00:13:08,140 --> 00:13:22,364 But given the structural constraints that we have now, I think the other thing to consider is that the legal industry is very much an apprenticeship industry versus other ones where 159 00:13:22,364 --> 00:13:29,528 there's a significant amount of spend in terms of the time that it takes to train juniors because they don't 160 00:13:29,528 --> 00:13:34,100 come out of law school knowing how to actually do the job itself. 161 00:13:34,100 --> 00:13:45,084 And so as we think about that sort of training process, there's great opportunity to insert innovation within that whole process where there's actually quite a big spend 162 00:13:45,084 --> 00:13:46,925 that's happening from law firms. 163 00:13:46,925 --> 00:13:57,009 So how does AI, how does innovation relate to that whole process where we're already spending in great amounts of time and resources? 164 00:13:57,229 --> 00:13:58,894 The other thing to think about 165 00:13:58,894 --> 00:14:11,874 is that although law school, or sorry, the legal industry is a trust industry, similar to financial institutions in that you don't really want your law firm going too far off the 166 00:14:11,874 --> 00:14:16,094 deep end and trying a whole bunch of risky things that maybe it'll work, maybe it won't. 167 00:14:16,094 --> 00:14:19,634 That's not really what clients want from a law firm. 168 00:14:19,914 --> 00:14:28,374 However, one of the advantages we have, and that allows, I think, the legal industry to move fast, even if we're not on the forefront, 169 00:14:28,374 --> 00:14:34,476 is that other industries are testing and trying a lot of the innovative technology. 170 00:14:34,476 --> 00:14:42,438 And the legal industry often is, you know, certainly not first to the party, but later on in that order. 171 00:14:42,438 --> 00:14:47,359 a lot of the risk in new technology has already been removed. 172 00:14:47,359 --> 00:14:52,871 And so that allows us to move a little bit faster while reducing that risk. 173 00:14:52,871 --> 00:14:57,858 So I think looking to other industries is really important in the legal industry because 174 00:14:57,858 --> 00:15:06,206 those are the ones that are taking those risks, failing in some cases, and then sort of testing what are the tools we can actually rely on. 175 00:15:06,587 --> 00:15:07,288 Yeah. 176 00:15:07,288 --> 00:15:16,074 Well, that's a good segue into what we were going to talk about in terms of AI and innovation and transformation. 177 00:15:17,356 --> 00:15:25,082 How do you see the role of AI in transforming knowledge management and library research functions? 178 00:15:25,082 --> 00:15:26,893 What's your perspective on that? 179 00:15:27,458 --> 00:15:40,888 Yeah, so I think there's two stages and this is not groundbreaking information here, but it's worth saying that the precursor to AI is really standardization and looking at what 180 00:15:40,888 --> 00:15:54,397 are the repeatable processes that we have, where is the consistent data that we can gather from those processes, where can we templatize things, these kinds of low tech activities 181 00:15:54,397 --> 00:15:55,916 are really the ones that are 182 00:15:55,916 --> 00:16:00,670 going to allow you to accelerate quite a bit when we get into AI. 183 00:16:00,670 --> 00:16:12,389 You you hear from either practitioners or just people in the market saying things like, let's just throw 300 documents into there and then say, now I want you to draft me 184 00:16:12,389 --> 00:16:16,702 something similar to that and make these changes. 185 00:16:17,263 --> 00:16:20,605 AI is just right now, it's not there yet, right? 186 00:16:20,605 --> 00:16:21,746 It's too... 187 00:16:22,104 --> 00:16:24,426 there's too much variance in terms of what you're going to get. 188 00:16:24,426 --> 00:16:30,219 It's not at the caliber of what someone would expect, especially of a large law firm. 189 00:16:30,480 --> 00:16:43,539 But if we take that extra step, if we do the spade work of really standardizing and looking at the repeatable processes and using knowledge management and research services 190 00:16:43,539 --> 00:16:50,508 and library services to get that standardized content, that is what's going to pour gasoline on 191 00:16:50,508 --> 00:16:56,166 what AI can do right now and we'll get a lot of efficiencies through that process. 192 00:16:56,668 --> 00:17:06,592 really the pathway to great disruption from AI is some of this low tech standardization work that needs to happen. 193 00:17:07,917 --> 00:17:11,489 IA before AI as they like to say, right? 194 00:17:12,270 --> 00:17:18,844 So yeah, at the KMNI conference last year, it's their second year. 195 00:17:18,844 --> 00:17:20,075 It's a great event. 196 00:17:20,075 --> 00:17:23,187 I learned so much from just sitting and listening. 197 00:17:23,187 --> 00:17:35,086 We sponsor, but I'm also an attendee and Sally Gonzalez from Fireman & Co did a, I think it might've been a keynote, one of the, maybe the first day keynote. 198 00:17:35,086 --> 00:17:37,687 And she outlined a framework. 199 00:17:37,815 --> 00:17:38,816 knowledge management. 200 00:17:38,816 --> 00:17:40,016 It was really good. 201 00:17:40,016 --> 00:17:42,817 Sally's been around since the beginning. 202 00:17:43,475 --> 00:17:50,435 I thought to myself, where does AI fit into that picture? 203 00:17:50,435 --> 00:17:54,552 I have some ideas on that, but I'm far removed. 204 00:17:54,552 --> 00:17:55,643 You're in the thick of it. 205 00:17:55,643 --> 00:18:04,707 How do you see law firms aligning like KM frameworks and the outcomes they want to achieve with AI? 206 00:18:05,966 --> 00:18:11,686 Yeah, so one treasure trove of opportunity is the document management system. 207 00:18:11,686 --> 00:18:22,345 And I've already described one way we can approach that is by looking at where's the activity in terms of the documents that are being generated, the documents that we're 208 00:18:22,345 --> 00:18:28,506 working on for clients, and looking at how can we standardize and templatize those things. 209 00:18:28,986 --> 00:18:34,444 Gowing really is a leader in that area of creating precedent material to really speed up 210 00:18:34,444 --> 00:18:35,174 efficiency. 211 00:18:35,174 --> 00:18:39,987 And so I'm really lucky coming into a firm like this, where we've got this huge head start. 212 00:18:39,987 --> 00:18:43,679 And I think AI is going to just accelerate that like crazy. 213 00:18:43,919 --> 00:18:55,275 The second area, though, is really around the data points that we can gather from the document management system from the structured information we have in our financial 214 00:18:55,275 --> 00:18:57,286 systems and that kind of thing. 215 00:18:57,306 --> 00:19:04,396 And it takes a little bit of curation and coordination in order to make sure that the data that 216 00:19:04,396 --> 00:19:10,438 these AI systems are going to sit on are cleaned up and are telling the right story and that kind of thing. 217 00:19:10,438 --> 00:19:15,166 I think knowledge management has a role to play in that curation process for sure. 218 00:19:16,027 --> 00:19:16,467 Yeah. 219 00:19:16,467 --> 00:19:18,848 Uh, in, in my experience. 220 00:19:18,848 --> 00:19:29,331 so between info dash and predecessor company Acrowire, we've worked with this number is bigger now, but a couple of years ago, it was like 110 AM law firms. 221 00:19:29,331 --> 00:19:35,733 So I've had a kind of a good cross-sectional view and the law firm DMS is typically a mess. 222 00:19:35,953 --> 00:19:43,685 And you know, there's all sorts of naming conventions, all sorts of different types of artifacts in it. 223 00:19:43,685 --> 00:19:46,237 There's multiple drafts. 224 00:19:46,237 --> 00:19:50,939 There's documents that were never used. 225 00:19:51,460 --> 00:20:01,105 It really feels like the curation, there needs to be some curation before we can leverage a DM in mass. 226 00:20:01,185 --> 00:20:04,140 But how do you, yeah. 227 00:20:04,140 --> 00:20:07,903 I think AI actually helps in that process as well. 228 00:20:08,004 --> 00:20:20,203 As we can deploy large language models on the document management system, which is something that every customer of a big DM is asking for either implicitly or explicitly to 229 00:20:20,203 --> 00:20:27,539 the vendors, as we can unleash AI broad scale across the DM and we know exactly what we're looking for. 230 00:20:27,539 --> 00:20:30,830 We know exactly how we want to tag certain types of documents. 231 00:20:30,830 --> 00:20:36,894 We know exactly the structured information that we want associated, like deal points and that kind of thing. 232 00:20:37,334 --> 00:20:40,957 Quite frankly, large language models can do that work. 233 00:20:40,957 --> 00:20:44,179 That works not that hard for the large language models to do. 234 00:20:44,179 --> 00:20:57,458 We just need a vendor to make those connections for us, to do that spade work of connecting the capabilities of large language models to this large scale of millions and 235 00:20:57,458 --> 00:21:00,930 millions of documents that can be under the direction 236 00:21:01,010 --> 00:21:06,738 of knowledge management professionals who will direct the AI to say, is what we're looking for. 237 00:21:06,738 --> 00:21:09,001 This is the structured information I need. 238 00:21:09,001 --> 00:21:16,200 And it's going to help that whole data cleaning and cleansing process quite a bit by leveraging AI on the front end. 239 00:21:16,683 --> 00:21:20,786 I'm of the opinion that it has to be the DM vendors that do that. 240 00:21:20,786 --> 00:21:40,650 And the reason is because of the scale of, and if you're in a cloud platform like I manage cloud or NetDocs, you can't interrogate a corpus of that size efficiently in any way. 241 00:21:40,650 --> 00:21:43,022 You have to be on top of it, right? 242 00:21:43,022 --> 00:21:45,113 Like infrastructure wise. 243 00:21:45,113 --> 00:21:53,227 It has to be very, very close, not over a WAN connection where the ingress and egress on that would be just a deal breaker. 244 00:21:53,227 --> 00:21:53,890 It'd be too slow. 245 00:21:53,890 --> 00:21:55,256 It'd be too expensive. 246 00:21:55,256 --> 00:21:57,298 It'd be very, very expensive. 247 00:21:57,298 --> 00:22:02,714 partially I understand why this hasn't been rolled out now, even though it's technically possible. 248 00:22:02,714 --> 00:22:08,630 We're waiting for prices to come down and more efficient reg pipelines in order to do it. 249 00:22:08,630 --> 00:22:18,720 But it's something I think everybody really needs and can leverage the document management system in a lot better ways once it happens. 250 00:22:18,961 --> 00:22:19,221 Yeah. 251 00:22:19,221 --> 00:22:20,282 And they're working on that. 252 00:22:20,282 --> 00:22:29,541 had Dan Hawk, the chief product officer from NetDocs on the podcast a couple months ago, and he described some ways that they're working on this. 253 00:22:29,541 --> 00:22:30,841 So they are. 254 00:22:31,663 --> 00:22:38,589 It's just, we're not there yet and it's still in flight. 255 00:22:38,589 --> 00:22:45,797 tell me about like, how should firms go about assessing the incremental value of 256 00:22:45,797 --> 00:22:49,008 these legal specific AI tools. 257 00:22:49,008 --> 00:22:49,908 There's so many. 258 00:22:49,908 --> 00:22:58,511 Like if I was in the CKO seat or the CINO seat, I would really be scratching my head how to tackle. 259 00:22:58,511 --> 00:23:02,112 There's so much out there and it's increasing every day. 260 00:23:02,112 --> 00:23:10,605 Like how should firms be thinking about finding the right solution for them and measuring incremental value? 261 00:23:11,995 --> 00:23:16,878 Well, I think first of all, firms need to actually ask that question. 262 00:23:17,378 --> 00:23:27,465 You know, just to sort of rewind a little bit, if you look at sort of the progression of the legal AI tools, which I think are wonderful and are very important, and I don't want 263 00:23:27,465 --> 00:23:29,546 to disparage them at all. 264 00:23:29,567 --> 00:23:36,571 I just really want to, I ask a lot of questions as a person, and so I think law firms should be asking a lot of questions as well. 265 00:23:36,571 --> 00:23:40,748 But if you look at the history, you know, we sort of went from zero 266 00:23:40,748 --> 00:23:44,341 to all of a sudden now we've got these legal AI tools, right? 267 00:23:44,341 --> 00:23:48,824 Like it happened really fast since the launch of the large language models, right? 268 00:23:48,824 --> 00:23:57,912 So a lot of people didn't have time to acclimate themselves in their personal life to making use of these large language models on a day-to-day basis. 269 00:23:57,912 --> 00:24:06,390 Early adopters like you and me were probably on chat, GPT and Claude and other tools right away, constantly using them, looking at the opportunities. 270 00:24:06,390 --> 00:24:10,871 in our own either personal lives or in a business and that kind of thing. 271 00:24:10,871 --> 00:24:19,303 And so then when we look at the legal large language models, model tools, you know, we're asking that question. 272 00:24:19,303 --> 00:24:27,546 So how is this incrementally better than what I could do in, you know, cloud or chat GPT or whatever, right? 273 00:24:27,546 --> 00:24:29,136 Like we're naturally asking that question. 274 00:24:29,136 --> 00:24:31,937 I think that law firms need to ask that question. 275 00:24:31,937 --> 00:24:36,268 And this is not to say that we would say, we don't need 276 00:24:36,418 --> 00:24:39,761 legal AI tools, I don't think that would be the conclusion. 277 00:24:39,761 --> 00:24:54,663 But it might be that there are certain use cases where we really do want to deploy a tool that has either training or they've nailed down the prompts under the hood to deal with 278 00:24:54,663 --> 00:24:56,314 legal use cases. 279 00:24:56,335 --> 00:24:59,878 you don't want to have to train lawyers on how to prompt effectively. 280 00:24:59,878 --> 00:25:03,761 You want it to be as natural as possible to the way they actually practice law. 281 00:25:03,761 --> 00:25:05,240 And the legal tools 282 00:25:05,240 --> 00:25:08,271 they really do help that in a lot of use cases. 283 00:25:08,451 --> 00:25:18,534 But then there are maybe other use cases where you might say, I don't know if there's an incremental value of the legal AI tool and maybe it would be better to have a secure 284 00:25:18,534 --> 00:25:25,055 version of the base model that is released within the firm for certain use cases. 285 00:25:25,356 --> 00:25:35,008 So I think these are important questions that need to be asked and every firm may come up with a different conclusion to that question. 286 00:25:35,205 --> 00:25:35,645 Yeah. 287 00:25:35,645 --> 00:25:41,100 Your point about experimenting on a personal basis is so incredibly important. 288 00:25:41,100 --> 00:25:50,938 Ethan Malik has a 10 hour challenge where he, you he challenges people to spend 10 hours performing tasks that they already do, but actually, you know, block time on your 289 00:25:50,938 --> 00:25:57,453 calendar, an hour every day for 10 days and run tasks through. 290 00:25:57,453 --> 00:26:01,677 Like a good example is how I prepare for this podcast. 291 00:26:01,677 --> 00:26:05,209 So I have a 30 minute planning call like you and I did. 292 00:26:05,443 --> 00:26:07,724 I download the transcript. 293 00:26:07,724 --> 00:26:11,416 have a custom Claude project with custom instructions. 294 00:26:11,416 --> 00:26:22,450 I've uploaded all sorts of KM specific material, like the skill last two years of the skills survey and like that outlines the priority because my audience is KM. 295 00:26:22,450 --> 00:26:28,613 want, I want my agenda, my agenda is to reflect the interests of the KM and I audience. 296 00:26:28,613 --> 00:26:33,703 Um, I uploaded examples of agendas that I wrote by hand. 297 00:26:33,703 --> 00:26:39,085 There's a couple of chapters from some books on knowledge management that I uploaded. 298 00:26:39,085 --> 00:26:53,812 And then I have custom instructions that tell this custom project how to go about its work, you know, to leverage the training materials first, to use a business casual tone, 299 00:26:54,212 --> 00:26:57,394 to align the topics to the training material. 300 00:26:57,394 --> 00:27:00,595 And I have taken, you know, how I used to do this was 301 00:27:00,667 --> 00:27:03,589 I would record the call and then go back and listen and take note. 302 00:27:03,589 --> 00:27:05,230 was ridiculous. 303 00:27:05,230 --> 00:27:16,815 Um, but I've saved so much time, but all of that process, like now I know how to build, I know how to build custom GPTs and Claude projects and I've incorporated them into my 304 00:27:16,815 --> 00:27:18,097 workflow. 305 00:27:18,097 --> 00:27:23,700 People need to understand how these things, and that's just, it's just a fancy rag implementation is all it is. 306 00:27:23,700 --> 00:27:26,021 Um, but I, I hear you, man. 307 00:27:26,021 --> 00:27:29,463 People have to spend time with the tools. 308 00:27:29,943 --> 00:27:35,426 and allocate resources towards that to really understand how it's going to apply in a work context. 309 00:27:35,426 --> 00:27:41,851 Yeah, and I would say that this is especially true of decision makers around technical purchases. 310 00:27:41,851 --> 00:27:47,036 I don't think the average person really is going to spend that time experimenting. 311 00:27:47,036 --> 00:27:51,979 I'm similar to you in that I use chat GPT all the time. 312 00:27:52,040 --> 00:27:59,606 And now that they've integrated with Siri where I don't have to have the voice mode open all the time to ask the one-off questions, you can just do it. 313 00:27:59,606 --> 00:28:02,168 It's even more the case. 314 00:28:02,168 --> 00:28:04,906 Literally just yesterday, I... 315 00:28:04,906 --> 00:28:10,479 you know, one use case for me in catching the train is I just want to know when the next trains are coming. 316 00:28:10,479 --> 00:28:13,198 And don't want to go through, open an app and this and that. 317 00:28:13,198 --> 00:28:14,320 And so I was sitting on the train. 318 00:28:14,320 --> 00:28:18,002 was like, why don't I just use generative AI here? 319 00:28:18,002 --> 00:28:23,484 Because I don't want to write the code that's going to go through a JSON object and figure out the times and that kind of thing. 320 00:28:23,484 --> 00:28:31,928 So I just opened the webpage, stuck it into a shortcut, and then just created a prompt, which you can do with shortcuts on the iPhone now. 321 00:28:31,928 --> 00:28:40,905 that says, look at this webpage, here are the data points that I want you to pull from it, and then tell me the next two trains, whether it's express and that kind of thing. 322 00:28:40,905 --> 00:28:46,729 But there's just so much low hanging fruit now that people can actually do. 323 00:28:46,830 --> 00:28:58,178 And we're in a totally different world that we have not yet seen the effects of these massive productivity gains across many different industries and many different areas. 324 00:28:58,683 --> 00:29:09,786 Yeah, we had a, we had a sales call yesterday with a firm who had their head in a place where they were going to do a custom dev project for their internet, which we did that for 325 00:29:09,786 --> 00:29:11,127 when we were known as Acrowire. 326 00:29:11,127 --> 00:29:12,867 We did that for like 12, 13 years. 327 00:29:12,867 --> 00:29:17,409 Like I know all the opportunities and challenges associated with that. 328 00:29:17,409 --> 00:29:21,430 And we decided to go down the product route because it's a better path. 329 00:29:21,470 --> 00:29:24,371 So, um, but right before the call, 330 00:29:24,571 --> 00:29:34,468 you know, I thought to myself, man, if I could use a metaphor here to describe, because you know, one of their concerns was, well, we don't know if, actually this was a, this was 331 00:29:34,468 --> 00:29:35,129 a different call. 332 00:29:35,129 --> 00:29:42,254 They, they wanted to, do a pilot and they weren't sure if their data was going to work with our tool. 333 00:29:42,254 --> 00:29:48,539 And I wanted to articulate to them that, you know, our solution is built for law firms and only law firms. 334 00:29:48,539 --> 00:29:50,030 And we've been doing it 16 years. 335 00:29:50,030 --> 00:29:52,155 So I got on chat GPT and advanced 336 00:29:52,155 --> 00:29:54,717 voice mode like three, four minutes before the call. 337 00:29:54,717 --> 00:30:04,445 And I said, Hey, I have a customer who is unsure about, um, how well their data is going to work with our solution. 338 00:30:04,445 --> 00:30:13,553 And I want to articulate a metaphor to help, you know, paint the picture and man, it nailed it. 339 00:30:13,553 --> 00:30:20,849 So it gave me a couple of great examples, like a, CUS a key custom cut for a lock that works every time a bespoke suit. 340 00:30:20,849 --> 00:30:22,401 gave me a whole bunch to choose from. 341 00:30:22,401 --> 00:30:26,526 And I was able to use that in my conversation. 342 00:30:26,526 --> 00:30:30,792 And it took me three, I started the process three, four minutes before the call. 343 00:30:30,792 --> 00:30:33,692 It's incredible what you can do these days. 344 00:30:33,692 --> 00:30:43,553 that's a great example for legal use cases actually, because what you and I are doing when we're using these models, we kind of have an understanding of what they can do and what 345 00:30:43,553 --> 00:30:44,574 they can't. 346 00:30:44,574 --> 00:30:53,403 And so we're able to guide and direct the tool to do what it's actually good at and then piece that together with the next component, right? 347 00:30:53,403 --> 00:30:56,766 This is sort of what I call the process of disaggregation. 348 00:30:56,896 --> 00:30:59,617 And that's what has to happen with the large language models. 349 00:30:59,617 --> 00:31:01,968 Now they can answer certain things, but not other things. 350 00:31:01,968 --> 00:31:05,449 They can process data and pull certain things out of it. 351 00:31:06,070 --> 00:31:08,110 Lawyers don't really think like that. 352 00:31:08,131 --> 00:31:20,356 And if, you think about for myself, having practiced at the junior level, there are some partners who, when they give you instructions, it's sort of like a high level thing. 353 00:31:20,356 --> 00:31:26,604 And then you've kind of got to go and figure out what does this even mean and start knocking on doors and asking people. 354 00:31:26,604 --> 00:31:31,566 And then there there's the rare partner that says, you know, these are the steps you're going to take. 355 00:31:31,566 --> 00:31:39,719 And after you get the result of that, then I want you to come back to me and we'll discuss that and then I'll send you off to the next piece. 356 00:31:39,719 --> 00:31:40,020 Right. 357 00:31:40,020 --> 00:31:52,265 So they're already disaggregating the overall task in their head so that a student isn't or a junior is not running off on rabbit trails, developing a completed project that makes 358 00:31:52,265 --> 00:31:55,336 no bloody sense whatsoever because they don't know what they're doing. 359 00:31:55,336 --> 00:31:56,096 Right. 360 00:31:56,364 --> 00:31:58,095 That's a large language model. 361 00:31:58,095 --> 00:32:04,339 We have to disaggregate the overall task and say, let's start with the creative process, right? 362 00:32:04,339 --> 00:32:07,221 I just need a metaphor and here are some examples, right? 363 00:32:07,221 --> 00:32:11,765 Then it comes back with some things and you say, hey, I really like the third one there. 364 00:32:11,765 --> 00:32:13,266 Let's work with that. 365 00:32:13,266 --> 00:32:17,128 And now the next step is I want to do A, B and C, right? 366 00:32:17,128 --> 00:32:21,431 Like that process of disaggregation and working iteratively with the tools. 367 00:32:21,431 --> 00:32:25,173 That is what I think is really going to change in the practice of law. 368 00:32:25,394 --> 00:32:26,284 But 369 00:32:26,518 --> 00:32:31,976 There's a learning process in terms of how lawyers are going to have to do that in the future. 370 00:32:32,327 --> 00:32:33,008 Yeah. 371 00:32:33,008 --> 00:32:41,914 Well, how, how should, um, you know, listeners of the podcast are probably sick of me talking about this, but it's, it's true. 372 00:32:41,914 --> 00:32:54,423 And so many firms are going a different direction, which is, know, when you're looking at defining an overall AI strategy for the firm, um, my, my recommendation is starting small 373 00:32:54,423 --> 00:32:56,804 and starting in the business of law. 374 00:32:56,945 --> 00:33:00,167 And the reason I think that makes the most sense. 375 00:33:00,167 --> 00:33:09,047 is first of all, business of law functions are the ones that are going to support the practice when the, when the practice of law solutions get deployed. 376 00:33:09,047 --> 00:33:10,867 So they need to have an understanding. 377 00:33:10,867 --> 00:33:13,607 So starting there is good for that. 378 00:33:13,607 --> 00:33:25,247 You don't have to worry about typically client data getting, you can use the public models for most of this stuff, like marketing, for example, a lot of times HR, like writing job 379 00:33:25,247 --> 00:33:29,847 descriptions, um, cut, you know, brainstorming comp plans. 380 00:33:30,147 --> 00:33:41,775 Um, finance, you know, there's all sorts of business of business of law functions that could, you could start small with and expand, um, incrementally. 381 00:33:41,775 --> 00:33:48,560 And I see a lot of firms not starting small, you know, they're jumping, you know, head first into Harvey or other tools. 382 00:33:48,560 --> 00:33:58,507 And, uh, you know, I, until recently, I knew very little about Harvey, but, uh, I had not seen a lot, but now they're starting to, and it's a, it's a great platform. 383 00:33:58,629 --> 00:34:05,244 I'm not saying don't do that, but I'm saying start small wherever you start. 384 00:34:05,244 --> 00:34:07,876 And business of law is not a bad place. 385 00:34:07,876 --> 00:34:09,567 What's your take on that? 386 00:34:09,742 --> 00:34:12,482 Okay, so I think there's two components here. 387 00:34:12,482 --> 00:34:16,422 One is start small, which I am 100 % on board with. 388 00:34:16,422 --> 00:34:28,342 And I'll sort of describe how I approach AI from a strategic standpoint, especially since I'm involved in talking about disruptive innovation across other industries in my teaching 389 00:34:28,342 --> 00:34:29,462 capacity. 390 00:34:29,782 --> 00:34:33,022 The second is the business of law component. 391 00:34:33,022 --> 00:34:36,182 So that's one way to start small for sure. 392 00:34:36,682 --> 00:34:39,922 The difficulty there, and I think you're right. 393 00:34:40,408 --> 00:34:47,113 But the difficulty there is when clients ask, how are you using AI to make things more efficient? 394 00:34:47,113 --> 00:34:50,466 And we say, oh, we're writing job descriptions with AI. 395 00:34:50,466 --> 00:34:54,129 like, oh, OK, really not that interesting kind of thing. 396 00:34:54,129 --> 00:35:06,319 So I think there's a pressure from the client side for sure for law firms to do the sexier, more innovative kind of uses of AI in the actual practice of law. 397 00:35:06,359 --> 00:35:10,412 But let's chat about that sort of start small. 398 00:35:10,890 --> 00:35:12,311 methodology. 399 00:35:12,351 --> 00:35:23,709 And, you know, when you look at the history of innovation and where disruptive technologies have come onto the scene, and there there's been winners and losers, some 400 00:35:23,709 --> 00:35:27,802 companies approached it really well, some companies didn't approach it so well. 401 00:35:27,802 --> 00:35:39,170 The ones that navigate disruptive technology really well, are the ones that look for immediate use cases of the technology that are very realistic. 402 00:35:39,170 --> 00:35:45,753 with respect to what the technology can actually do right then in the short term. 403 00:35:46,554 --> 00:35:57,120 Rather than trying to boil the ocean and, sometimes you win, sometimes you lose with that approach, but they look for those incremental innovations with the new technology. 404 00:35:57,120 --> 00:36:04,644 So an example that I use sometimes is when inkjet printers came out. 405 00:36:04,725 --> 00:36:07,042 Companies that navigated that really well, 406 00:36:07,042 --> 00:36:16,535 were ones where although they had perhaps a stranglehold over the laser printer market, they looked at this disruptive technology that was not good enough for their core 407 00:36:16,535 --> 00:36:17,545 customers. 408 00:36:17,545 --> 00:36:27,148 And so they didn't go out and say, okay, let's start pushing inkjet printers on all of our core customers and disrupt ourselves and that kind of thing. 409 00:36:27,328 --> 00:36:35,202 They decided let's create a new product line here that can operate independently and we can experiment. 410 00:36:35,202 --> 00:36:44,769 with this new technology and we'll let it grow and see where it goes rather than sort of sitting back, sticking our heads in the sands, letting other smaller companies begin using 411 00:36:44,769 --> 00:36:50,512 the disruptive technology and perhaps displacing us at some point. 412 00:36:50,533 --> 00:36:52,554 So let's keep it internal. 413 00:36:52,554 --> 00:36:54,208 Let's use it for what it's good at. 414 00:36:54,208 --> 00:36:58,879 This is not our core customer, but let's find the customers where this could actually help. 415 00:36:58,879 --> 00:37:04,402 And I think what law firms need to do now is look at when we're talking about AI in the practice of law, 416 00:37:04,406 --> 00:37:08,869 Certainly the business of law, there's tons of opportunities and I'm 100 % on board. 417 00:37:08,869 --> 00:37:10,770 We need to be doing that. 418 00:37:10,810 --> 00:37:19,905 But as we're looking at the practice of law, what we don't wanna do is throw AI at complex legal tasks that it's actually not very good at because lawyers are going to be very 419 00:37:19,905 --> 00:37:26,940 frustrated with trying to use a tool for something that it should not be used for right now. 420 00:37:26,940 --> 00:37:30,988 We need to look at those low level tasks. 421 00:37:30,988 --> 00:37:35,220 that lawyers are involved in in the process of producing legal work. 422 00:37:35,220 --> 00:37:47,797 Things like document review and the creative process in drafting and where you've got a narrow universe of content that you are plugging into the LLM to ground it in that 423 00:37:47,797 --> 00:37:50,609 content, whether it's a contract or a document. 424 00:37:50,609 --> 00:37:54,211 And there are very specific things that you are asking about it. 425 00:37:54,211 --> 00:38:00,014 Those are the use cases that generative AI is quite good at right now. 426 00:38:00,242 --> 00:38:03,946 And it's not everything that lawyers do, but we need to be realistic about that. 427 00:38:03,946 --> 00:38:09,010 Let's focus it in on what the AI is actually good at right now. 428 00:38:09,627 --> 00:38:10,248 Yeah. 429 00:38:10,248 --> 00:38:21,163 You know, trial prep is another use case that I think AI is outstanding for, you know, in a litigation scenario, you know, witness preparation. 430 00:38:21,222 --> 00:38:22,653 Yeah, what am I missing? 431 00:38:22,653 --> 00:38:26,196 Are there any contradictions in what the witness previously said? 432 00:38:26,196 --> 00:38:27,276 Those kinds of things. 433 00:38:27,276 --> 00:38:29,915 It's really, really useful for those things. 434 00:38:29,915 --> 00:38:33,047 Yeah, yeah, that's kind of low hanging fruit on the practice side. 435 00:38:33,047 --> 00:38:42,682 So you know what's interesting about your comment and I agree with you and I understand like RFPs coming out asking firms how they're using these technologies. 436 00:38:43,623 --> 00:38:47,155 Starting small should have started a year ago, right? 437 00:38:47,155 --> 00:38:51,867 So that that and obviously we can't go back in time. 438 00:38:51,867 --> 00:38:57,540 I understand that, but it's OK to start small aggressively, right? 439 00:38:57,540 --> 00:38:58,390 So. 440 00:38:58,919 --> 00:39:05,023 start small in the areas where you have the highest risk reward equation balance, right? 441 00:39:05,023 --> 00:39:06,664 That's most favorable. 442 00:39:06,804 --> 00:39:14,169 And in the practice of law, the use cases that we've been kicking around here are great examples. 443 00:39:14,470 --> 00:39:28,643 But you know what's funny is like you see this disparity, this disconnect, like some firms, and I think this has died down over time, but some firms in their OCGs required 444 00:39:28,643 --> 00:39:33,115 law firms to agree not to use the generative AI on their matters. 445 00:39:33,115 --> 00:39:36,356 And then you have other firms that are saying, how are you using it? 446 00:39:36,356 --> 00:39:36,956 You know what I mean? 447 00:39:36,956 --> 00:39:38,627 Like you need to be more efficient. 448 00:39:38,627 --> 00:39:44,638 there's been, I think over time that's that it's kind of like cloud, right? 449 00:39:44,638 --> 00:39:58,775 Like cloud, you know, um, it's exactly like cloud, you know, it's it, only thing is the clients aren't demanding that firms use cloud, but it was clients that held firms back. 450 00:39:58,929 --> 00:39:59,839 for the most part. 451 00:39:59,839 --> 00:40:09,177 And just there's, I've talked to some folks who had, you know, really kind of a skewed picture of the value prop, but yeah, it's like we. 452 00:40:09,223 --> 00:40:19,432 like in organizations, right, you hear the headlines of whatever company has seen proprietary code from a competitor and that kind of thing through using generative AI. 453 00:40:19,432 --> 00:40:29,140 And then that filters into these freak out sessions where, you know, we sort of think about what, what do we need to do to eliminate this risk with people who don't understand 454 00:40:29,140 --> 00:40:31,722 what the technology actually is or what it does. 455 00:40:31,722 --> 00:40:37,016 And then that goes through multiple levels where now that's been institutionalized and these are 456 00:40:37,016 --> 00:40:42,091 various guidelines and that kind of thing, which are for the most part, we're seeing less and less of it. 457 00:40:42,252 --> 00:40:48,118 However, I think what's always important with a new technology is to step back. 458 00:40:48,118 --> 00:40:51,041 And this is what I did three years ago at my previous firm. 459 00:40:51,041 --> 00:40:56,807 The question I kept asking is in five years, are we going to care about this? 460 00:40:56,807 --> 00:41:00,406 Is anyone going to care about these questions in five years? 461 00:41:00,406 --> 00:41:04,588 And that's not to say that, okay, so let's plow ahead and do it anyway. 462 00:41:04,588 --> 00:41:05,419 Absolutely not. 463 00:41:05,419 --> 00:41:07,000 You respect what the client wants. 464 00:41:07,000 --> 00:41:08,951 That's first and foremost. 465 00:41:08,951 --> 00:41:21,417 But as we're thinking strategically, as you're building your AI strategy within the organization, you have to have a long term enough vision that you're not building things 466 00:41:21,438 --> 00:41:28,001 and moving forward with initiatives that are just not going to matter in the medium term. 467 00:41:28,222 --> 00:41:28,602 Right? 468 00:41:28,602 --> 00:41:29,402 So 469 00:41:29,420 --> 00:41:32,871 I think that's really important as we think about strategy. 470 00:41:32,871 --> 00:41:45,015 And we've seen examples where people have looked at what can an LLM do now and why don't we train this on our data and we'll build our own LLM and do this and that, right? 471 00:41:45,015 --> 00:41:52,417 And then the new version of the model comes out and it blows away what everybody has been working on for the past year internally, right? 472 00:41:52,417 --> 00:41:56,792 It's again, because of a myopic vision of 473 00:41:56,792 --> 00:41:58,443 where the technology's at right now. 474 00:41:58,443 --> 00:42:08,830 You've got to at least look out five years in advance with any billed project or any significant process and think about where will the technology be then? 475 00:42:08,830 --> 00:42:10,121 What will people care about? 476 00:42:10,121 --> 00:42:13,113 How will we be practicing law at that point? 477 00:42:13,113 --> 00:42:21,218 And then how do we build our capacities now in order to prepare us for what law is going to look like in five and 10 years? 478 00:42:21,873 --> 00:42:23,443 makes perfect sense. 479 00:42:23,584 --> 00:42:37,984 And, you know, I haven't had a chance to, I'm not willing to write the $200 a month check to kick the tires on, uh, three yet, but I'm hearing from multiple sources that it, it 480 00:42:37,984 --> 00:42:40,768 really is incredible what it can do. 481 00:42:40,768 --> 00:42:49,141 And if you listen to Sam Altman and many others, they, there's a, there's a differing definitions on, on AGI and what that means. 482 00:42:49,141 --> 00:42:51,291 Uh, I don't know that there's really, 483 00:42:51,417 --> 00:42:56,468 any agreement, any any semblance of agreement on what age AGI is. 484 00:42:56,548 --> 00:43:13,213 But, um, what I, what I have heard is AGI that are a definition that makes sense to me is AGI is when AI can perform the overwhelming majority of tasks better than any human across 485 00:43:13,213 --> 00:43:14,413 any discipline. 486 00:43:14,413 --> 00:43:14,753 Right. 487 00:43:14,753 --> 00:43:18,354 That's, that sounds like a reasonable definition of AGI to me. 488 00:43:18,354 --> 00:43:21,297 What will that mean when we get there, whether it's 489 00:43:21,297 --> 00:43:25,652 You listen to Sam Altman and it's going to be the end of this year or you listen to others. 490 00:43:26,835 --> 00:43:28,167 you know, it'll be three years from now. 491 00:43:28,167 --> 00:43:33,183 What is that going to mean in a legal context and the practice of law? 492 00:43:33,210 --> 00:43:34,671 yeah, that's that's a question. 493 00:43:34,671 --> 00:43:37,092 I don't think I'm qualified to answer that. 494 00:43:37,092 --> 00:43:39,373 That's that's the question that we're asking, though. 495 00:43:39,373 --> 00:43:46,216 Like, what is law going to look like in a world where AI can do everything that a lawyer can just as well? 496 00:43:46,216 --> 00:43:53,069 And I think there's a couple of things that we're going to see in that kind of a world, assuming that world materializes. 497 00:43:53,069 --> 00:44:01,708 You know, as you're thinking about strategic planning, you plan for the most basic probabilities of what could happen. 498 00:44:01,708 --> 00:44:07,581 And certainly there's a possible world where what you just described actually happens, right? 499 00:44:07,581 --> 00:44:13,144 AI can do law just as well, if not better than people can. 500 00:44:13,144 --> 00:44:16,826 And we have to plan for that possibility. 501 00:44:17,166 --> 00:44:22,049 And I think in that kind of a world, the value proposition of a lawyer is very different. 502 00:44:22,049 --> 00:44:31,574 You know, the types of people that we want to hire are not necessarily the folks that are going to sit behind a computer screen and do technical areas of the law. 503 00:44:32,283 --> 00:44:35,986 what we look for in terms of grades and that kind of thing. 504 00:44:35,986 --> 00:44:47,686 It may change in terms of what we're looking for in lawyers coming out of law school because there's going to be a lot more complex, multidisciplinary problems where lawyers 505 00:44:47,686 --> 00:44:58,085 function a lot more like consultants, using AI to do the technical stuff and then solving these business problems, these complex business problems for their clients. 506 00:44:58,149 --> 00:44:59,039 Yeah. 507 00:44:59,039 --> 00:45:02,012 Legal today is so work product focused, right? 508 00:45:02,012 --> 00:45:03,283 That's the deliverable. 509 00:45:03,283 --> 00:45:12,650 And the reality is that lawyers have so much more to offer than just a word or a PDF document. 510 00:45:12,650 --> 00:45:13,771 It's their knowledge. 511 00:45:13,771 --> 00:45:14,671 It's their context. 512 00:45:14,671 --> 00:45:15,632 It's their understanding. 513 00:45:15,632 --> 00:45:17,714 It's their ability to connect dots. 514 00:45:17,714 --> 00:45:25,563 It's their ability to project in the future, knowing like from a regulatory perspective, what's in queue and what's likely to come. 515 00:45:25,563 --> 00:45:29,635 to fruition with the new administration, for example, like that sort of stuff. 516 00:45:29,635 --> 00:45:30,917 AI can't do that. 517 00:45:30,917 --> 00:45:31,147 Right. 518 00:45:31,147 --> 00:45:40,835 Because there's reading of tea leaves that has to happen there and extrapolation, and it would be very difficult, um, to pull all those pieces together. 519 00:45:40,835 --> 00:45:41,596 So I agree with you. 520 00:45:41,596 --> 00:45:48,802 It's going to turn more into a consulting business than just a pure, you know, work product deliverable scenario. 521 00:45:48,802 --> 00:45:50,863 Like it happens a lot today. 522 00:45:51,246 --> 00:45:53,626 And I think there's a couple other things as well. 523 00:45:53,626 --> 00:45:57,906 So one is the role of a lawyer is definitely going to change. 524 00:45:57,906 --> 00:46:07,965 The other thing I think law firms, especially large firms, are going to be tasked with handling mass, absolutely astronomical amounts of data. 525 00:46:08,046 --> 00:46:16,146 If you think about what AI enables in our legal system, Our legal system, the sort of the pinnacle of it is the courts, right? 526 00:46:16,146 --> 00:46:19,926 That's sort of what drives everything where decisions are made. 527 00:46:20,342 --> 00:46:25,367 Right now there is an artificial barrier to entry into the whole legal system. 528 00:46:25,367 --> 00:46:32,634 It's quite expensive to hire a lawyer and I've got a dispute or whatever it is. 529 00:46:32,855 --> 00:46:39,842 Once AI can do all of that stuff as well as a person, there's no more barrier to entry anymore. 530 00:46:39,842 --> 00:46:44,066 Our legal system is going to be completely flooded with cases. 531 00:46:44,318 --> 00:46:53,132 large corporate clients are going to be flooded with these cases that are coming in and they're going to need law firms with the technical capabilities to be able to handle 532 00:46:53,132 --> 00:46:55,183 massive amounts of data. 533 00:46:55,183 --> 00:47:05,447 So not only is the legal system going to be flooded for that, but even on the enforcement side, enforcement bodies are going to be able to use AI to find infractions that right now 534 00:47:05,447 --> 00:47:10,949 would take months and months and months of combing through paperwork in order to find the infraction. 535 00:47:10,949 --> 00:47:12,706 They'll be able to find it instantly. 536 00:47:12,706 --> 00:47:14,176 the risk of clients is going up. 537 00:47:14,176 --> 00:47:18,451 So there's just going to be a ton of work for law firms to actually navigate. 538 00:47:18,451 --> 00:47:28,679 So you've got this multidisciplinary consulting happening with this increase in volume, like a huge increase in the volume of work. 539 00:47:28,679 --> 00:47:36,165 So I think there are going to be different ways that law firms as almost a consulting organization are going to be able to deal with that. 540 00:47:36,326 --> 00:47:42,094 One is sort of the personal consulting approach, but the other one might be 541 00:47:42,094 --> 00:47:54,931 productized offerings that, you know, clients can use instantly in order to help them navigate things that leverage the data that law firms are going to accumulate through this 542 00:47:54,931 --> 00:47:56,402 increase in legal work. 543 00:47:56,635 --> 00:47:57,535 Yeah. 544 00:47:57,556 --> 00:48:09,170 And I know we're almost out of time, but one thing I wanted to quickly ask you about, which you alluded to in your intro was lessons from other industries. 545 00:48:09,170 --> 00:48:16,077 Like what can we learn from how other industries have managed disruption and innovation effectively? 546 00:48:16,077 --> 00:48:17,879 What are your, what are your thoughts on that? 547 00:48:18,082 --> 00:48:20,514 Yeah, I touched on this a little bit previously. 548 00:48:20,514 --> 00:48:29,252 So I think the starting point is to be realistic about where the technology is now, but also very cognizant of where the technology is going. 549 00:48:29,252 --> 00:48:33,145 And we've gone, you know, both these places in our conversation, right? 550 00:48:33,145 --> 00:48:39,901 Like AI, it's not doing the high value stuff right now, but there's a possible world where it does do that. 551 00:48:39,901 --> 00:48:46,586 So then the question is strategically for an organization, how do I leverage what it can do now? 552 00:48:46,742 --> 00:48:51,424 and build a blueprint for where the technology is going. 553 00:48:51,424 --> 00:48:58,527 So an example that I think I and probably others use all the time is that of Netflix, right? 554 00:48:58,527 --> 00:49:03,969 When Netflix built its company, it was on version one of the internet. 555 00:49:03,969 --> 00:49:07,180 There was no streaming, broadband wasn't happening. 556 00:49:07,180 --> 00:49:14,413 Like the technical capabilities were not there for an on-demand video delivery service, right? 557 00:49:14,693 --> 00:49:16,482 But strategically, 558 00:49:16,482 --> 00:49:22,264 They built their company to win in that world that was very quickly coming. 559 00:49:22,365 --> 00:49:25,726 So what do you need to win in a streaming world? 560 00:49:25,726 --> 00:49:26,986 You need a customer base. 561 00:49:26,986 --> 00:49:29,117 You need algorithms for recommendation. 562 00:49:29,117 --> 00:49:36,781 You need to get people used to clicking on websites in order to navigate their video rentals rather than going into a store. 563 00:49:36,781 --> 00:49:38,621 You need all these different capacities. 564 00:49:38,621 --> 00:49:40,232 You need an inventory catalog. 565 00:49:40,232 --> 00:49:43,133 You need licensing agreements with different companies. 566 00:49:43,133 --> 00:49:45,474 All this stuff is necessary. 567 00:49:45,518 --> 00:49:47,159 for the streaming world. 568 00:49:47,419 --> 00:49:48,380 But what did they do? 569 00:49:48,380 --> 00:49:53,943 They built a mail order business because that's what this technology could support at the time. 570 00:49:53,943 --> 00:50:02,848 And it wasn't as good as brick and mortar in some ways, but they made use of what the technology could do in the present in order to win the future. 571 00:50:02,888 --> 00:50:08,831 And that's really why I think it's important that law firms look at what can AI do right now? 572 00:50:08,831 --> 00:50:13,934 And how can we use that as a blueprint in order to win a future? 573 00:50:13,954 --> 00:50:17,437 where AI can lawyer just as well as people can. 574 00:50:18,011 --> 00:50:18,361 Yeah. 575 00:50:18,361 --> 00:50:29,166 And you know, what's another interesting aspect of Netflix and what they were able to accomplish is they started streaming other people's content and then they, they saw the 576 00:50:29,166 --> 00:50:41,641 writing on the wall of, you know, Disney and NBC and all these content producers creating their own networks and they got ahead of that and started doing original content. 577 00:50:41,641 --> 00:50:46,853 So yeah, kudos to Netflix for, you know, seeing 578 00:50:46,853 --> 00:50:49,525 the future and preparing for it proactively. 579 00:50:49,525 --> 00:50:59,974 They didn't wait until these content providers turn the screws on the licensing agreements to make their business unviable. 580 00:50:59,974 --> 00:51:06,849 They started early and got ahead of it and have been wildly successful as a result of that. 581 00:51:07,190 --> 00:51:09,201 It's what every successful company has done. 582 00:51:09,201 --> 00:51:14,664 You look at Uber, they didn't build their company in order to create a gig economy. 583 00:51:14,664 --> 00:51:16,395 They happened to do that. 584 00:51:16,395 --> 00:51:22,819 They built their company because of self-driving vehicles that is the future we're moving into, right? 585 00:51:22,819 --> 00:51:24,730 That's where the profitability is, right? 586 00:51:24,730 --> 00:51:27,662 So you look to the future, where are we going? 587 00:51:27,662 --> 00:51:31,384 But what do I need to build in the present in order to win that future? 588 00:51:31,384 --> 00:51:35,266 Because I need a customer base in order to win. 589 00:51:35,266 --> 00:51:39,043 the self-driving fleet possible future, right? 590 00:51:39,043 --> 00:51:44,252 So what do law firms need to build right now to win the future of AI lawyering? 591 00:51:46,299 --> 00:51:48,860 Yeah, that's a great way to end it. 592 00:51:48,860 --> 00:52:04,564 Those are very powerful analogies and ways to think about how law firms need to start thinking foundationally different like Uber and Netflix because that level of 593 00:52:04,564 --> 00:52:11,916 transformation is coming and the firms that prepare for it will be successful. 594 00:52:11,916 --> 00:52:15,597 Those that don't will be the blockbusters of the world. 595 00:52:17,570 --> 00:52:18,687 Very true. 596 00:52:19,316 --> 00:52:22,137 well, it's been a great conversation. 597 00:52:22,137 --> 00:52:25,118 We've been, again, talking about this forever. 598 00:52:25,118 --> 00:52:36,613 You transitioned roles, so I wanted to give you some time to get your feet underneath you there, but we really look forward to working with you at Gowling, and I appreciate the 599 00:52:36,613 --> 00:52:37,623 time today. 600 00:52:37,634 --> 00:52:39,711 Yeah, thanks Ted, thanks for having me. 601 00:52:40,953 --> 00:52:41,689 Absolutely. 602 00:52:41,689 --> 00:52:42,091 All right. 603 00:52:42,091 --> 00:52:43,094 Have a good one. 604 00:52:43,310 --> 00:52:44,169 You too. -->