[00:00:00] Bjarne, how are you this afternoon? Hey, really great, Ted. Thanks for inviting me. Yeah. I'm excited about the conversation. I saw you on a podcast, um, couple of months ago, and you had, you had some spicy takes and, um, I thought, "Yep, I gotta get, I gotta get Bjarna on the, on the pod." So I appreciate you being here with me today No, it's a pleasure. Looking forward to the conversation. Yeah. Well, um, before we jump into the agenda, maybe you can tell us a l-- You had a real storied career on the inside legal, um, in the inside legal world. Maybe you can tell us a little bit about your background and, um, what you're, uh, what you're up to these days. Sure. Um, so very briefly, I've had a 30-year legal career. I started out back in the mid-'90s, um, at White & Case and then at Sullivan & Cromwell in different parts of the world, starting in New York and then Stockholm, Helsinki, Frankfurt. Um, uh, [00:01:00] at, at which point I, I jumped in-house, um, first at Kimberly-Clark and then, uh, over to the Coca-Cola system, which was a client of mine from my law firm days and, uh, uh, went with Coca-Cola, uh, both on the bottling and on the company side, uh, again, all over the world, starting in London and then Vienna, Athens, Greece, uh, Tokyo, um, Atlanta, and then back to New York 20 years after I left, um, uh, practice there. And then jumped over to, uh, Pearson, which is a large FTSE 100 company. That was my first kind of, uh, public, uh, company general counsel role, and, uh, it was a trial by fire. There were a lot of reorganizations that needed to happen because the industry that Pearson was in, which was, uh, education, publishing, and media, so we owned assets like the Financial Times and The Economist and Penguin Random [00:02:00] House. They were all changing, um, and that was a great incubator for learning about, uh, process optimization, team technology, sort of in the very early days, uh, of the alternative legal service provider revolution. This was in 2014. Uh, and I spent, uh, quite a few years at Pearson and then, uh, moved over to GlaxoSmithKline, a large pharma company where I was part of the team that took, um, Haleon public. Haleon is the world's largest consumer health business. It has brands like Sensodyne and Advil and Theraflu and Tums and so on. Um, and that was an opportunity to kind of build the other way. So rather than kind of restructuring and downsizing, it was an opportunity to shape things moving up, uh, so building a team from scratch. And we were sort of in the, uh, foothills of the AI revolution, which kind of made that an interesting, uh, call. And [00:03:00] today, I'm the CEO of a, uh, an advisory business called Fjordstream Advisors. Uh, I'm sort of living the dream at the moment. Um, I, I get to tell other people what I think and they pay me for that, and that's very, very cool. Uh, I also do coaching, so I coach a lot of lawyers. Uh, usually lawyers in new roles, lawyers in transition, lawyers retiring. Um, I find that to be a very, uh, fruitful endeavor. And then I'm very engaged in academics and writing. Uh, I'm-- I teach at Yale. Uh, I'm a senior fellow at the London School of Economics, and I just wrote my second book called "Law in the Era of AI: Clients, Firms, and the Future of the Legal Industry." So that's kind of me in a nutshell, I guess. Well, that is very ti-- those are all very timely topics and, you know, your experience on both the buy side and the sell side of external legal services is super interesting because we really are at an interesting [00:04:00] crossroads in the industry, and it's something that we talk a lot about here on the podcast and that I write a lot about. I go to about 14, 15 conferences a year and speak at many of them, but also just listen and, and learn. And, um, yeah, we are, um, we are at a real pivot point with, with law firms. You know, there really has not been a technology transformation in legal that has impacted th-th-the foundation, right? We've had email, digital research, e-discovery. These things were, um, uh, ancillary might be, uh, too strong of a word, but AI is bringing a completely new dimension to every aspect of the business, especially the client service model, um, and how law firms service their clients, how they compensate their, uh, [00:05:00] associates and partners, the partner track, um, how they engage with clients, and all of that needs to change quite dramatically. And I, I'm not sure if y-your take on this, but it feels like we're off to a bit of a slow start, um, on the law firm side. I, I don't know if you saw, um, I pulled the article up here. Um, Stephen Crowley, the GC at Ford, posted an article in Bloomberg Law saying, "Hey, law firms, we're ahead of you on this tech transformation. We need you to catch up, and the ones that, that are all in on this are gonna get rewarded, and the ones who limp in, um, are going to get penalized." So I don't-- Is it your read as well that there seems we, we seem to be moving a little bit slower on the external side than internal buyers would like? Oh, I mean, uh, you know, definitely s- slower. Um, I don't think that's... Y- you know, and, and bas- actually, just [00:06:00] kind of one reaction to what you said a minute ago, um, it does-- I do agree with you. It does feel like all the innovations that have come up until this point have really been faster horse and buggies, and AI is the car, right? And, and so we are now, uh, in an era with a new general purpose technology, uh, very much like the steam engine or electricity, uh, that is, uh, highly prevalent, that will spawn, you know, innovations across every part of the economy, that will profoundly transform not just how law is practiced, uh, but, but absolutely everything else in our lives. And, um, a- a- a- and the other thing is I think law firms really aren't reacting any differently to the way every dominant incumbent in virtually every industry has reacted to profound technological change. So there's nothing [00:07:00] Kind of surprising about the fact that law firms want to protect their moat and have highly evolved structures that are perfectly attuned for the current operating environment. And so why would you and how could you disrupt the very thing that gives you so much, uh, profit? That's a challenge that every, you know, every industry faces. I think the, the one sort of add on to that is, and we can get into this, is that there are certain dynamics, uh, that relate to law firms, how they're structured, how they operate, their cultures that make it exceptionally difficult, right? Um, but you know, the fundamental proposition is very much there. And then to your point about the, um, you know, the article that you mentioned, yeah, I think we're living, you know, we are living in exponential times, but we think linearly, and we, we kind of work our way through the world in a linear fashion. And that means we don't often feel the ground moving under our [00:08:00] feet. Uh, and um, it's both, you know, terrifying and exhilarating. Uh, but you know, we wake up and one day it seems like everything has changed in a very, very short period of time, and that's kind of the world that we live in. Um, and I also fundamentally agree with the point that clients are changing faster, and we can talk about that too. I think, you know, if you ask me what is really gonna disrupt, uh, the legal profession, I wouldn't be as focused on any specific tool or technology or any player. Uh, I think it's the client, and I think the client is gonna disrupt it because the client is spending, you know, corporations are spending two point six trillion dollars this year on AI and AI related purchases, uh, and profoundly changing how they operate. And that is going to change legal departments and that in turn is going to change law firms. Um, and in my book, I actually go back and look at [00:09:00] that throughout history and sort of ask myself, has that ever happened before? That as changes in the macroeconomic environment happen, that changes companies, and as companies change, so do law-- legal departments. And as that shifts, so does the power differential and the dynamics between outside suppliers and buyers. Uh, and it turns out that is in fact the case Yeah. There was a, a, a recent report, it was from Deloitte last month, that had some really interesting statistics. I just posted on it this morning. Um, so Deloitte interviewed 121 legal in-house leaders and some interesting stats. Um, globally, the expectation is that the hourly rate work will fall from 72% to 44% within two to three years, right? So almost in half. In the US, it, it's expected to move from 82% to 56%, so [00:10:00] massive, you know, 60%, um, decrease. And that, uh, it says some GCs are committing to cutting external legal spend by 20% to 40% over the next three years. And that is going to have-- that is gonna directly pa- impact outside counsel revenue. So the question that I think a lot of people in the industry are grappling with is, will Jivan's paradox fill the void? Um, you know, will either latent demand or new demand, right? 'Cause there's a dif- a few different ways that we could fill that gap. Um, it could be demand that already exists that doesn't get serviced, or it could be just the complexity increasing of the w- of the world as a result of this technology and all of the regulatory implications and the legal work associated with that could also create new demand. I don't know. How do you see these scales balancing out [00:11:00] in, in, in the future? I mean, I do think, um, if you, you know, again, if you have history as a guide and you kind of look back and, uh, what you see is that as technology has, uh, been introduced that is of a general purpose nature, and as that technology transforms the economy, jobs, tasks, everything profoundly changes. Like if you went back to 17, you know, 76 and, uh, when the Constitution was, you know, being, uh, negotiated and the Declaration of Independence happened, 98% of people were engaged in agriculture. And if you had told, you know, Thomas Jefferson or, or whatever at the time that, you know, it by tw- 2026, um, 2% of the population would be engaged in agriculture, but they'd be producing 10 times as much food. Um, I think a lot of people at that time would be tempted [00:12:00] to panic and say, "Well, what are all those people gonna do? I mean, that's great that we have all this prosperity," but turns out, you know, we do all kinds of things. Um, the only difference is that this is all happening in months and maybe, maybe years rather than, you know, centuries, and, uh, that does introduce a challenge for us. I do think I'm, I'm very much an optimist in that respect, though. I do think, um, it-- there are gonna be a plethora of new roles, and I can already see them coming. You know, um, if you look at-- I mean, this is something we could also talk about. I think this disruption isn't just gonna wash over law firms, it's gonna wash over alternative legal service providers, and I think you're going to see a shift away from You know, SaaS models, a shift away from, uh, models that rely on, uh, process optimization and human input towards ALSPs that are developing new [00:13:00] technology, uh, innovations that will, for example, make agentic systems or generative systems more reliable, uh, that will improve, uh, governance guardrails, that will enable structured velocity. So, you know, how do you make sure that speed can happen at scale in a safe way? Um, all of these things have yet to be developed, and I think, uh, you know, if you just kind of look at what problem are we trying to solve for in the foothills, uh, of, of this transformation, a lot of the people I talk to are like, "This technology is great, but, um, frankly, I can't trust it, right? I don't know whether the output's a hundred percent..." You know, and we're in, in law, right? It's like I, I, I use it, and then I gotta g- I gotta put people on this to figure out whether the output is actually real or hallucinated. I don't trust the agents enough. I don't have enough orchestration [00:14:00] layers, and I don't have enough guardrails that would allow me to just put teams and teams of agents at work, right? And if you look at the capability frontier, this stuff is moving at an insane pace. Um, uh, I was looking at some, uh, some data from Meter the other day that said, uh, the task capabilities of frontier models is moving at four times the speed of Moore's Law, right? So Moore's Law, you know, the, the doubling every couple of years of computing power, effectively the number of microchips on a, on a, on a, uh, the number of transistors on a microchip, um, happening roughly every two years. Uh, task accretion, so the ability of these agentic frontier models to complete tasks, um, that would take humans a certain amount of time, is doubling every seven months. So, you know, two years ago, uh, frontier models could do the equivalent [00:15:00] of what it would take a human a couple of minutes. Last year it was an hour. This year it's a alm-- we're almost at a day. Um, next year maybe we're at a week or six months or a year. Uh, who knows? Uh, but that obviously has profound implications, but it does create all the problems we just talked about. How do you ensure governance? How do you make sure this stuff, um, actually doesn't go off the rails? And if, if you wanna talk about problems that we gotta solve for, that's-- there are enough jobs right there. Uh, I could imagine 50 new companies and, and the companies that are going to capture the most value are the ones that truly solve for those problems. Um, the other thing is we don't really understand who's gonna end up a winner and who's gonna end up a loser in this evolving ecosystem, and there's gonna be a plethora of players coming around, uh, all the layers. You know, whether it's the, the LLMs, whether it's the wraparounds and wrappers that sit on top [00:16:00] of that, whether it's, uh, various other enablers like platform models that sit between companies and law firms. So plenty of opportunity to imagine a whole range of new jobs and increase productivity over the next, uh, decade. Yeah, and the amount of-- You know, I've been in the legal tech space for almost 20 years, and, um, prior to maybe two years ago, m-maybe three, the amount of Silicon Valley VC attention in legal tech was, was minimal, and, uh, for good reason. I mean, l- um, you know, the law has not really been a very tech-enabled industry relative to others. Like I spent 10 years at, at, in banking at Bank of America doing process engineering and risk management work, and, you know, the transformation that happened in financial services was extraordinary. I mean, you know, mobile [00:17:00] banking, the, um Uh, the physical footprint of these institutions like Bank of America shrinking and in favor of digital channels, um, you know, um, ap-approvals of loans, like with a click of a button. It's been, it's been extraordinary what has happened in other industries, and like legal has been like one of the last holdouts, right, from this transformation. I don't think it was anything intentional, but, um, you know, the large language models that are based on language is the, is, is the basis of, of legal work. And now we have a technology that's finally really well aligned to disruption, and we're in a, we're in a place where, you know, to your point, we've got a ton of VC and PE money coming in and deploying capital, and we've got all sorts of, of, of new [00:18:00] needs like, uh, you know, the current paradigm. I always-- I think about this transformation in waves, and I need to formalize this into a post or an article, but I feel like we're in the first wave right now, which is, I'll call it the co-pilot era, not Microsoft Co-Pilot, but just you have an existing workflow, you have an AI tool, and y-you do things a little more efficiently. And I really see wave two as being where we re-architect the workflow and, um, and leverage systems more deliberately rather than, than bolting on. Um, and then finally, I think the, the third era or wave will be, you know, the, the agentic era or the, um, autopilot era where things happen. I've-- We had a, a guest on the podcast a few weeks ago, Grace Chediak from Google, and she was talking about some of the things they're experimenting with. Um, she's on the M&A [00:19:00] team and, you know, they are experimenting with agentic models to get to the zone of possible agreement agentically. So, you know, um, you really need agents on both sides to make that process efficient. But, um You know, we're, we're, we're starting to dabble in-- But reliability is still, to your point, a question mark, and, and we have a lot of, uh, shifting, um, uh, shifting landscape around what's this gonna cost, right? Like the all-you-can-eat model seems to be, you know, these token subsidies are, are starting to evaporate, and now firms are being left to foot a really large bill. I know, you know, I, I read your LinkedIn, and you, you talk a lot about, uh, agentic systems. Um, it, it feels like we're really early in, in this, and I don't see a lot of things in production. I see more [00:20:00] experimentation, but what, what are you seeing? No, I, I agree 100%. Um, I think y-you put your finger on a number of sort of layers that I think the next wave, um, will, will be focused on. We talked about governance already. I mean, the other one is cost, right, and token, token usage. Um, sort of how do we-- As you begin to scale these things is wh- what I'm seeing when I speak to people who work inside large corporates, uh, who are doing things like, uh, what you've explained at Google, sort of using LLMs, um, the kind of law version of L- of some of the LLMs, the large ones, um, uh, to negotiate against, against themselves, create digital twins, um, uh, you know, come up with basically scenarios for your negotiations playing out over, you know, the entire length of that, uh, period and creating alternate models before you even go in the [00:21:00] door. All of those things are very cool, um, but they also cost a huge, uh, amount of token usage and a large amount of energy. And so that's another problem that, that needs to be solved, right? And, you know, agentic, I think, is one way to do that. I mean, one of the things that's interesting here is, you know, at the moment, if you look at some of the layers that sit on top of those LLMs, like the Harveys and the Lagoras I'm not so sure, you know, users have full ability to modulate between expensive frontier model defaults and cheap, let's say cheaper low energy using models, right? Uh, you do if you're using Anthropic, you can kinda select the model you want and the level of intelligence you need, uh, and you can kind of coordinate that in a way that keeps things cost-effective for you and manages your, um, you know, your allocated amount [00:22:00] of usage. You don't necessarily have that same flexibility with a lot of the wrappers. Um, I, I think at the moment they do it behind the scenes. So, you know, they're kind of... I-- M-maybe there's an arbitrage going on there. I just, I, I don't know enough about it. Um, and maybe I'm totally wrong and, you know, Harvey will complain and say, "No, no, no, you've got full control over this." But my, my, my sense is there could be more control there over what LLM we're using for what task. You know, I can imagine tools coming up there to help modulate that more effectively across an enterprise. I mean, there are so many places where there are bottlenecks and there are needs, and there are ways that you could imagine streamlining costs and improving service delivery in an AI era. Um, so just, just because, you know, yesterday's approaches to this in a pre-AI era, uh, aren't [00:23:00] necessarily that effective now, it doesn't mean, uh, there won't be a whole slew of new businesses and use cases that start to emerge over the next six to 12 months. Yeah. I, I ran into an int-interesting statistic the other day too, speaking of efficiency, and, you know, I've heard people talk about, um, you know, one mitigating factor to the token cost situation that we're finding ourselves in is efficiency. And yes, we can absolutely get more efficient, but, um, the stat I saw was around what percentage of the pop-population actually uses AI. In the US, it's less than one in four, right? So less than twenty-five percent. Globally, it's one in five or one in six. So if you think about the amount of demand that hasn't yet come online What-- Like efficiency, yes, but capacity has to be [00:24:00] a, a really big part of that conversation, and not just compute, um, you know, energy. It's-- There, it's-- There are many constraints in the system that, that have to be addressed. Efficiency isn't gonna solve them all because we're just beginning to bring this demand online. So I'm, I'm hopeful, and I fully expect, I mean, all technical systems get more efficient over time. Um, but I do think that we are still gonna run into some serious constraints I, I think you're right. You know, although I think necessity is a mother of invention . And, you know, if you think about, let's say, fuel efficiency in vehicles, um, you know, as, as more and more people began using vehicles and as vehicles, as cars became, you know, a critical part of modern infrastructure, fuel efficiency, you know, was a problem that needed to be solved, [00:25:00] right? And today, you know, modern cars, uh, use less gasoline than a car in the 1970s used parked with the engine off. Uh, there was, you know, um, th- so there, there are like surprising, shockingly surprising statistics at how, um, there was a need for greater efficiency, and that need got plugged by innovators, um, as the market matured. Um, so I agree with you. I think, um, the other thing that's interesting when we think about efficiency, and we've talked a little bit about, you know, can you create tools that allow users to automate modulation between frontier models and lower cost models and different levels of energy, uh, intensive intelligence based on the task. That's something that we currently have to do manually if we have the opportunity to do that at all as customers. Um, you [00:26:00] also have questions around sort of where inside the architecture of a legal department is energy and efficiency most critical, right? Not everyone uses AI tools the same way. You might have someone, uh, sitting in one part of the department that's barely s- you know, scraping the surface of token usage, uh, the allocated token usage, and then you might have someone else building, you know, advanced models, uh, in the legal ops team that is, you know, f- far in excess of the, the allocated, uh, token usage for the department per head, right? And so again, you have opportunities to sort of target that, um, and come up with more effective, uh, models. Um, I think also, you know, we haven't talked about open source, who's gonna win, um, as we go forward. I think you're gonna end up with premium layers and value layers [00:27:00] a-across the market. But, you know, some of the open source LLMs, the Chinese ones, for example, that are, um Probably significantly, you know, certainly cheaper, uh, maybe somewhat less reliable than frontier models. And y-y... But, you know, uh, y-y-you can download those, you can adjust the weights yourself. Um, it's not a subscription. Y-you know, y-y-you can lower your costs by mixing and matching the tools and the models that you need for the use cases that you're trying to address. So I, I just think there are a lot of opportunities here, and, and you're, you're absolutely right. We're in the foothills of usage across, uh, you know, consumers and indeed across c-companies. I, I don't think every company is sort of, um, you know, at AI factories status level with digital platforms at the heart of everything they do. I think most companies are [00:28:00] still feeling their way. But the amount of spend that's going into this, uh, is mind-boggling, and, uh, it's-- I think it's gonna profoundly change how companies operate. And I'm already seeing it, you know, in, uh, in quite a few non-traditional, non... Sorry, non-digital companies, uh, that have moved in that direction and achieved, uh, incredible scale and efficiency by putting digital at the heart of their operating models. You know, and it's interesting you talked about mother or, uh, necessity being the mother of invention. Like, we've really imposed that paradigm on China, um, with our export controls, and look what's happened. You know, we've got new architectures like mi-mixture of experts where you have, you know, Kimi two point three trillion parameter model that lights up a small fraction based on the input. And, um, we are [00:29:00] teaching China how to be more efficient because they have to. So I completely agree with you there. Um, one thing that I, I read in your writing that I want-- I'd love to get your, uh, explanation about is the gazellephant, um, the AI factory or client and the rise of the gazellephant. Tell, tell me what, what that means. So a gazellephant, um, you know, if you think about, uh, maybe, maybe the way to start that conversation is to-- The gazellephant is a metaphor for what, uh, a legal department inside a digitally enabled AI-centric company, uh, looks like structurally. And, uh, t- b-before describing what the gazellephant is, it's probably useful just to take a second to, to reflect on what is actually happening inside companies. Like, what, what do companies look like, um, structurally as they put these AI-centric models, um, at the core of their [00:30:00] operation? And, uh, you know, in short, it's what Marco Iansiti and Karim Lakhani at Harvard Business School have coined, uh, the AI factory, um, which really is a metaphor-- It's ob-- It's, it's more of a, it's more of... It's less a metaphor and more of an actual model, a, a description of what these models look like. But the idea is data comes into the company from all sorts of places. It gets organized and cleansed and sorted and tagged in ways that allow machines to read it. Uh, a-and by the way, there's a huge opportunity there around data also in legal, if we're talking about needs in the, in the AI era, that is a massive one. I mean, the, the teams that can come in and organize people's data is, you know, are gonna, are gonna mint it if they can do that well. But the data comes into companies, it gets sorted and organized, and then it gets fed into these AI systems, right? Algorithms, agentic models. Uh, and those [00:31:00] systems generate insights, they generate recommendations, and in some cases, they drive business outcomes. And in all of those actions that come out of this, uh, model, you have data that gets fed back in, right? Fed back into the machine, so you have these continuous learning loops, which means that you can increasingly automate what you do, and you can speed it up at machine speed and scale. Um, so, you know, an obvious example that we're all familiar with is, you know, when you call an Uber And millions of people are calling Ubers like every day. Uh, th-there aren't like human dispatchers who decide what car comes to your house, right? That's all happening automatically, right? The search ad auctions that happen on Google millions of times a day are run completely automatically. There is no human being, uh, that runs those. And as that speed and scale picks up inside a company, as that engine begins to [00:32:00] turn, you generate enormous value, right? Walmart, uh, digitized about five years, four or five years ago during COVID. It now has a valuation of, uh, over a trillion dollars in market cap, right? It has a P/E ratio that exceeds many of the hyp-hyperscalers. Um, that, that is what happens when companies adopt these digital, uh, platforms for a whole range of reasons. So if you're a legal department inside one of these, uh, machines, you're now suddenly faced with a client that is generating not one version or five versions of your next marketing plan, uh, for a particular product, but it can generate, you know, 400 versions, and it needs turnaround in 24 hours, not in two weeks. And so there's almost like an arms race going on inside these companies where you start to feel the heat. Now, in the old days, like a year ago, two years ago, it was, it was all about, you know- The old [00:33:00] days. Right. In the old days, you know, uh, Richard Susskind wrote a lot about, you know, the more for less challenge, and I, I did and lots of other people did. And that challenge was, was all about sort of reducing your resources and at the same time increasing your output, um, which, you know, was mainly an exercise in cutting people and maybe, uh, enlisting basic technology and process optimization. Today, the more side of the more for less is really where survivability and success lie, right? You've got to be able to invest in these new tools in the same way that your company is. You have to put an, a mini AI factory at the heart of your legal department so that you can kind of generate the same scale, scope, and learning effects that are happening inside the, uh, the business. That's the idea. And then the gazellephant is kind of like, well, what does that look like? Um, and years ago I-- when I was [00:34:00] at Coca-Cola, I traveled across rural India with the president of the Indian operation, and, uh, we were, we were going through all these places where, you know, there were new consumers who were making a few dollars a day coming online as consumers. And he said to me, "You know, to win in this fast-moving emergent market, you need to be an elephant on the back end," meaning you've gotta have a stable supply chain, and you've gotta be a gazelle on the front end, so meaning you've gotta have a sales force that is hugely nimble and able to kind of react and adapt to local needs and be creative in terms of how you sell, uh, into markets where people are making three or four dollars a day, right? Um, and I came back home after that trip, and I was telling my team about this during a brainstorming session. Uh, uh, and, uh, my secretary was like, "It's a gazellephant." Um, and we started thinking about this, and we're like, "Yeah, that's actually a great metaphor for an AI [00:35:00] era legal department." It's a stable back end that's a tech platform enabled, uh, stably, you know, stable digitally, uh, you know, generating outputs and insights that get, uh, pushed into the department that allow the department to punch above its weight. Um, and then at the front end, at the business end, you have small teams of highly skilled specialists and generalists who can move much more quickly because now they're enabled by, uh, the AI inputs. And, and all the while, that is generating new insights and data that get fed back into the platform, generating a continuous loop of learning and creating the same kind of scale, scope, and learning effects you see inside the business. So that's sort of the idea Wow. So, um, that sounds like a tall order. I mean, because there's, there's tension, right, between the gazelle and the elephant, right? The stable back end [00:36:00] creates rigor, structure, process, um, friction that the-- is gonna slow the gazelle down. Having worked in organizations of all size, I've seen it, I've seen it firsthand. Um, h-how-- i-is it, is it a realistic objective, um, to achieve something like that? Uh, what, what, what are your thoughts? So gazelle-phant models exist in every company, right? Uh, virtually every company has a stable, uh, supply chain, for example, uh, that provides reliability where it matters most and a nimble emergent front end that moves with-- at the speed of the market. Uh, and I think you're already seeing this, and some of these things predate, um, sort of what we might call the modern AI era. They go back sort of five, six, seven years. But if you look at, say, JPMorgan, uh, JPMorgan [00:37:00] Chase put in place, um, a system called Coin quite a few years ago now. Uh, it's a loan agreement tool, uh, that, uh, is AI-enabled. Uh, and in the first year alone, that tool Uh, was capable of replacing three hundred and sixty thousand hours of human time. Um, it can process twelve thousand, uh, documents a second, uh, and it has an error rate of one percent, um, which means that the bank saved a hundred and fifty million dollars in fraud losses in the first year alone, um, because it's better at predicting fraud than humans. Um, that's scale inside a legal department. Um, scope inside a legal department is, uh, you know, Honeywell, I think, is an example of a company that increasingly is pushing its legal department data into the enterprise, [00:38:00] um, and combining that data with enterprise-wide data like HR data, compliance data, finance data, in ways that allow the business as a, as a whole to move more nimbly. Uh, and then you have continuous learning loops happening inside contract management systems that self-adjust playbooks based on prior successful deals, fraud models that self-adjust based on, uh, input on the latest, uh, scams out there, um, s- and so on and so forth. So yes, I think it is, uh, doable, and we're seeing the, the manifestation of the front end, uh, a-across a number of businesses. What's required to do that is process optimization and data organization, and those two things, you know, are a lot harder than they seem. And, um, you know, a few years back, I wrote a, an article together with Dan Wu and Michele DeStefano called, uh, [00:39:00] "Don't Let the Digital Tail Wag the Transformation Dog." And I think the title says it all, but the idea is As GCs, as general counsel, you know, think about technology, they often start at the wrong end because they go, "Gee, wow, we need to be gazellephant. That means, you know, we gotta go out there and acquire like some AI technology. So let's go trawl through whatever vendors have hit our systems in the last, you know, and, and let's pick some cool looking AI and let's, you know, let's get a great dashboard up and..." And that's kinda like going to the, you know, going to Williams Sonoma on the weekend and getting seduced by the pasta maker and then buying it, and it takes up expensive real estate in your kitchen, and you never use it. Why? Because it's not solving for a problem you ever had, right? And the better way, the real way to start this process is begin with the elephant side of the gazellephant. So begin with what is my [00:40:00] purpose? What is my unique selling proposition? Why do I exist? Why do I have a right to get paid every month? And, and what are the problems I'm trying to solve for that are preventing me from being fully able to deliver on that purpose? Now I understand what problem I'm solving for, now I can begin designing my organization. I can begin building an org model around that problem. And only then should I begin thinking about what technologies might accelerate that redesign, right? So it's a-- it, it ha-- And then you've gotta navigate through the change element of that, like the human cost of navigating all of this. So this is not an easy lift. And, and then I would add, you know, in the years since that paper was written, data, data, data. Like you've just got, you've got to organize your data as part of that stable backend. You are not gonna be able to put an AI tool on top of data that is, you know, disconnected sitting in people's, uh, [00:41:00] desktop folders that is, you know, only partially accurate, uh, you're gonna get very confident output that is very wrong, right? Or, or that sits in silos across the enterprise where it can't connect and it can't speak to, uh, they can't speak to the-- to, to each other. If you don't have your data set up, then the front end is not possible. So y- it's all about structure, form, process, data, purpose, understanding what problems you're solving for, putting that layer down, then you can move it at speed and scale. And, you know, maybe a, a last analogy I've used in other contexts, like in a governance context, but I think it, it, it's also apropos here is, you know, I live in Germany, and Germany has, uh... One of the coolest features, uh, of Germany is that it has, uh, the Autobahn, right? It's, uh, it's basically long stretches of highway where there are no speed limits. Um, you can drive as fast as your car, uh, takes you and as fast [00:42:00] as your skill will allow. Um, but that only works because you have infrastructure, right? You've got, uh, you have, you know, lanes that are designed, you know, in the curves to enable a car to come at 250 kilometers an hour and not blow out, right? You have lane changing norms. You do not drive in the fast lane. Uh, you just pass very quickly and go back. You do not pass somebody on the right side in a slow lane. Um, if a speed limit comes on because it's raining, you follow that speed limit to the letter, and if you violate any of those things, you're gonna lose your license in a second, right? And that is what enables that speed to happen. So it's really the same thing. It's infrastructure that's laid down, its rules, its norms, and those things enable acceleration to happen on the front end. You know, it's been years since I've worked in an enterprise environment, but when I did, I was-- before BofA, I was at Microsoft, um [00:43:00] I had a different picture of legal and I, and I, I worked with legal closely in some of those years. Uh, what-- The picture you painted was very integrated, right? Um, you know, data flowing in and out l- in lockstep with the business' needs, and that's not, that wasn't my experience. Um, my experience was very siloed, firewall. We had a portal that we used to engage. I had a couple of patents that, um, I had to work with the legal team on. I was in anti-money laundering, and my listeners have heard me talk about when we would, you know, identify, um, cases where legal had to get involved. It was very siloed, firewalled. Even the reporting structure was completely separate from the business. And what you're talking about here sounds very integrated and sounds infinitely better. Um, I'm just... Is, is it, is it happening like that? Is this, um, [00:44:00] aspirational, uh, or are inside legal teams really delivering on this? So I think it's a great question. I think, um, uh, my experience to date has been that there is a, uh, there is a, uh, frontier group of companies that are way ahead and that are doing incredibly cool things, and they tend to be attached to companies that have accelerated, uh, the, the business as a whole i-into this space. And then you have a very long tail of companies where, you know, the GCs have no idea wh-- how to even begin this whole process, which is what keeps me in business, right? And, um, uh, e-examples of, you know, the former, um, uh, you know, one example that I, I would give, that I give in my book is Workday. Um, what Rich Sauer and Anya Lyons at Workday are putting together is [00:45:00] almost a blow-by-blow the Gazellephant model that I articulated. Uh, and, um, uh, including, you know, they're now working on embedding instead of just having a digital front door where people come when they need, um, uh, work, you know, from legal, they have agents embedded in the enterprise. So if you're in Salesforce or if you're in SAP and you're working on something, uh, one of the agents, the personas from legal may recognize that you need legal support, uh, and will s-send a, a message to the triage agent, which in turn will activate a process. So, uh, those are companies that are out on the front end. There are also companies doing interesting things with technology process optimization. Unilever is o-- another one, um, uh, that are building, um, operation centers where they're consolidating their technology [00:46:00] footprint and professionalizing process optimization so that the front end, which is separated and embedded in the business, can operate at scale. So you have businesses that are moving, uh, at, uh, great speed, and then you've got that long tail of companies and legal departments that are still struggling. And then you have everything in between. I'd say it's no different from business. I mean, businesses are all over the map on this, and it just comes down to how well run is your company. Uh, that makes a big difference. But even in a poorly run company, I would submit that if you are an enlightened general counsel and you have a desire to move ahead, you can be a pioneer in this space and you can do things. And you don't need a 1,000 person legal department to do this. I, I would say it's actually easier if you're like a 10 person legal department to optimize your processes and put some baseline technology in place on top of that, that will move things and allow you to operate [00:47:00] with way more efficiency and scale, uh, than your number would submit, right? Interesting. Yeah, if there's, if there's executive support and resources necessary in, in a, in a organization like that, absolutely. Yeah, you have less, you have less cultural momentum to pivot, um, in a small organization. We're almost out of time, but I wanted to ask you about, uh, one thing on the agenda that we were gonna touch on was timeline. So I-- The way I think about the timeline of how this transformation unfolds and the disruption that's gonna take place in the law firm world, to me feels like a very slowly then suddenly dynamic, right? And I think what's, what is driving that slowly th- then suddenly, um, dynamic is [00:48:00] that so much of what law firms are doing today is happening in isolation. Another stat from that Deloitte report that was just a complete eye-opener is that o- f- fifty percent of legal departments report only an initial conversation with their panels about AI, and twenty percent report no conversation at all. So that means that there's, what? Seventy percent that are barely engaging with their clients on how to develop this new client service model, pricing model, all the things that need to happen. And what it feels like is the law firms are doing it in a vacuum. The GCs are moving full steam ahead, and there's not tight engagement. It's very loosely coupled. So y- in terms of, do you kind of see a, a very [00:49:00] slowly then suddenly scenario, or do you have a different take on that? No, no, I mean, I, I agree 100%. Um, I can't remember who said this, but somebody said, "Smoke, snow, snow melts from the edges." Uh, and, uh, you know, I think it's a perfect analogy for, um, many of the theories of disruption. If you think about Clayton Christensen and some of the writings that he, he made, um, "The Innovator's Dilemma" being maybe the most prominent, uh, book in that category. But he really spent a lot of time studying disruptive innovation and change, and the pattern is typically that change starts at the periphery, right? It begins at the low end of the market. Um, you have barbarians at the gate. You have entrants that come in at that low end. They Perhaps initially, um, they have early versions that are imperfect. The technology or the solution that they're providing, um, is done differently, [00:50:00] and it appeals to maybe a different segment of the market. And so the incumbents, in this case, the law firms, you know, tend to dismiss that, right? They focus on their core business, they focus on the main value proposition. And the thing is that, you know, over time, gradually, gradually, gradually that those in-- those disruptors, the technologies that they bring, the new solutions get better and better and better, and then suddenly they're good enough. And the minute they're good enough, everyone wants that. Um, you know, and I think a great example from the business world, right? Nokia, uh, the world's largest handphone, cell phone maker, right, that made handsets, uh, back in, I think it was two thousand and seven, the same year the iPhone was launched. They were on the cover of, um, Fortune, I believe it was, uh, with the cover said: "Can, can anyone stop the cell phone king? One billion sold," or whatever it was. And, um, that same year, you know, the iPhone came out. [00:51:00] Nokia dismissed them. They're like, "Oh, you know, they don't have keyboards. They're expensive. They appeal to a niche market. They're never gonna c-capture the business segment." And, um, you know, what they missed was not only did, you know, the iPhone get better and better, but it actually changed the entire game, right? Because now it was no longer about, uh, the cell phone, it was about the App Store, it was about iOS, it was about, uh, all the things you could do with a phone. I-- now suddenly I had all these new jobs I wanted to have done, right? I wanna, I wanna be able to take a picture of my dinner and upload it in- onto Instagram, right? That's something I now wanna do that I never wanted to do before. And you can have the world's best handset, but if you can't do that, it's like having the world's fastest horse and buggy. It was great until suddenly the car was good enough, and then it was like, "Whoa, I want a car." And if you ask your customers, they'll tell you they want more of what [00:52:00] you're giving them until suddenly they don't Um, and that's-- so that's partly why that's such a deceptive, difficult thing for an incumbent to see. And by the way, all the while that this is happening under the surface, you're making record profits, right? Um, you've had the best year ever. So, you know, being able to go into a partnership, sort of, uh, a partner, you know, profit distribution meeting and tell your fellow partners, "Hey guys and gals, I have an op-- I have a suggestion. Why don't we, uh, why don't we take some of the money we were going to distribute to ourselves and divert it, um, to invest in disruptive technology solutions that will blow up this profitable model in five or 10 years when none of us are around? Who's for that?" Right? Like, nobody, right? So you have that dynamic happening. Um, I will say there's another piece to this that's interesting because I think one of the questions I get all the time is like, "Well, when is this gonna happen?" Right? [00:53:00] Um, uh, Rita McGrath, who, um, is a Columbia Business School professor and wrote a, a book called, I think it was called Seeing Around Corners. Absolutely great book. She, she kind of took that whole disruption theory and looked at, are there signals in the noise that we can use to understand when change is imminent? And she looked across different industries, and she calls these inflection points. And a couple of the ones that she identified included, uh, low emp-employee morale, right? Uh, cheaper, good enough solutions. Customers are no longer excited or maybe even hate certain features about the, the current offering. Um, there are unexpected new entrants and predictions of change. And when you kind of look at that and you look at the current environment that we're in, right, where I saw an Axiom survey from 2024 that said 100% of the 250 plus large cap GCs that were interviewed [00:54:00] said that cost, quality, and other challenges made them regret law firm engagements, right? 89% of them said they no longer view law firms as a completely effective solution. 96% were facing budget cuts. So you kind of, you kind of look at that and you have to ask yourself You know, are the signals telling us something or are we just completely unique and immune from the laws of every other industry and nothing is going to change and we can keep going as before? Now, I think the challenge for law firm leaders is how do you do something with all that, right? I mean, what are you going to do if you're running a highly profitable, perfectly evolved organizational model, um, where you don't have absolute control? It's a coalition of the willing with partners, and firms can suddenly collapse, right? So you have all of these dynamics that make it super, super challenging. And yet what I will say is [00:55:00] I do work with some law firms. Some of them are doing amazing things, and they are tearing pages out of the playbook of the innovator's dilemma, and they're investing in little small experiments, and they're, they're learning from those things. Uh, and you know, so there are ways that you can do this well, but it is not easy. And, um, you know, I don't envy, uh, m-managing partners of firms that, uh, you know, have the vision and understand what needs to happen, but it, it's kind of like being, uh, in a, in a place where it's difficult to move and you see a train coming. You know, it's, it's, uh, it's not an easy environment. The, the deck is really stacked against big law. I mean, the ABA rules are at the top of that list. You know, uh, Model Rule 54 doesn't allow external capital, uh, or fee sharing except through messy workarounds like the MSO. You know, you've got the, uh, the highly portable partners who can take their book of business. There's no [00:56:00] non-competes. Um, you've got the billable hour that is the axis around which the law firm revolves and needs a complete rethinking, like lawyer identity that's very tightly coupled to billable hours for compensation and just i-i-- there's just so much cultural inertia that has to be redirected. And to your point, you've got people at the top of that seniority stack that are the most influential with the shortest retirement horizons that are gonna be asked to make investments that they'll never see the payoff on I, I think that's right. And yet at the same time, I, I, I, I'm, I'm not a complete pessimist. I, I actually believe some law firms are not only gonna survive but actually thrive in the AI era. Uh, but it's not gonna be the firms that stand still. It's gonna be the firms that really step back and take a very strategic look at what they do that [00:57:00] continues to endure and that, that clients will continue to pay for in a world where, you know, expertise is increasingly commodified, um, and judgment, uh, you know, and judgment and wisdom are what people are prepared to pay for, right? Um, and I can imagine a whole range of different models emerging. I think some will kind of be like Savile Row tailors. Um, you know, there are still these fancy tailors in London that will give you a £10,000 suit handmade in a world where Zara and H&M are, are prevalent. Um, they're, they're just not very scalable, but they're highly profitable, and I think there will be some firms that will, will manage to migrate there. I think you'll have firms that kind of operate as hybrids, like Swiss Army knives that have legal services, but increasingly will offer like a whole range of other services like technology, uh, consulting, risk advisory. [00:58:00] Uh, so kind of like a miniaturized big four for law. Uh, I think, uh, outside the US certainly you're seeing already the beginnings of corporate firms with, you know, uh, private equity coming in, backing firms, maybe ultimately with a view of taking them public. There are some firms in the UK that have done very well under that model. Keystone is an interesting, uh, example of that. And then in the US I think, you know, you, you might even see some very interesting PE back moves, um, you know, what you might call a capital consolidator, uh, uh, model where You take everything but the legal advisory and you suck it out of the law firm, uh, and you turn that into a back-end service organization that charges a fee for technology, billing, HR, like everything. Uh, and now you've got an incentive to bring those costs way, way down because your margin goes up. Uh, and if you can do that successfully with [00:59:00] one firm, you could scale that across every firm in the United States, and you could become a massive business. So, you know, where is the real money? Um, I, I think there are ways to do this, and I think, you know, coming back to where we started, this era that we're in is one of the most exciting eras in history. It's terrifying, uh, but it's also exciting, and I think there will be tons of opportunities, um, to make money, to have great jobs, to reinvent and reimagine how legal services get delivered, and to do that in a way that gives access to legal services to more people who need it, uh, at a lower cost and with greater predictability. So I'm-- Ultimately, I'm a, I'm an optimist, even though I recognize it's gonna be bumpy before we get there. Completely agree. I'm, I'm in for it. I've, I've oscillated between utopian, dystopian, um, you know, SaaSpocalypse, and I've-- I'm here for the whole ride, but overall, I'm, I'm, I'm, [01:00:00] I'm bullish. Uh, well, this has been a great conversation. I apologize I, I kept you a little bit longer than I said I would. It was a, it was a fantastic, uh, conversation, and I really appreciate you taking some time with us today. Um, before we go, how do people find out what's your new book? Where do they buy it? All that sort of stuff. Yeah. No, thanks. So first of all, thanks for having me. Really enjoyed the conversation. Uh, the book is called Law in the Era of AI. You can find it, uh, everywhere books are sold, Amazon. Uh, it's published by Wiley, so you can get it, um, you know, on, on the internet, in bookstores. Uh, and you can find out more about me, um, I, you know, I'm on LinkedIn, bjarnetallman.com, and also Fjordstream Advisors is the name of my company, and we have a website. Uh, I recognize Fjordstream is difficult for many non-Norwegians to pronounce or spell. It's F-J-O-R-D-S-T-R-E-A-M [01:01:00] advisors.com, all one word. Uh, but, uh, there you have it. And listen, thanks for, for having me. I'll look forward to following, uh, future episodes and your writings and, um, and to staying in touch. Absolutely. All right. Thank you very much, Bjorn. We'll, uh, we'll chat soon. Hey, thanks a lot. Stay well. All right, take care. Thanks for listening to Legal Innovation Spotlight. If you found value in this chat, hit the subscribe button to be notified when we release new episodes. We'd also really appreciate it if you could take a moment to rate us and leave us a review wherever you're listening right now. Your feedback helps us provide you with top-notch content