Irene Liu: What happens when one of the world’s largest law firms decides that AI is not simply another technology to buy, but a capability it needs to build for itself? Welcome to AI Sidebar. I’m Irene Liu, your host and executive director of the AI Initiative at Stanford Law School. Today, I’m joined by Erica Berthou, a member of Kirkland & Ellis’s Global Executive Committee and co-head of Global Fundraising and Fund Transactions, who has helped lead the firm’s AI initiative from the beginning.

Kirkland recently announced plans to invest more than 500 million in AI and technology over five years. But the more interesting question may be what that investment says about where the legal industry is headed. After evaluating thousands of AI products, Kirkland chose to work with Palantir to build proprietary technology around the firm’s own knowledge, and importantly, the judgment of its lawyers. That raises some much bigger questions. Can a law firm create a real competitive moat with AI? Does greater efficiency finally put pressure on the billable hour? And what happens to firms that do not have the scale to make investments like this? Erica and I get into all of it, so let’s jump into the conversation.

Hi Erica, thank you so much for joining us on the AI Sidebar.

Erica Berthou: Thank you. It’s super great to be here, Irene.

Irene Liu: So Erica, one of the most fascinating things of legal tech news in the summer was Kirkland’s announcement that it plans to invest more than $500 million in AI, and that’s over a span of five years. So Erica, how did Kirkland and your team decide to calibrate the size of that investment at such a large amount, and what gave you confidence that this was the right moment to make such a large bet?

Erica Berthou: Sure. It is a large amount indeed, and it is super exciting. We are obviously all energized over here about it. For us, it wasn’t really a question of how much we should spend on AI per se. It w- the question really was, what would it take to fundamentally rethink how we work in an AI first world? I mean, we’re obviously a huge institutions with thousands of lawyers, an enormous amount of data, is, like, so much knowledge moving around the, the firm every single day.

So at our scale, treating AI as a side project would have been super risky, obviously. So if you believe that technology is gonna reshape how work gets done, and that’s our belief, you don’t just optimize around the edges. You go all in. So you build the infrastructure, bring in the talent, and give yourself the ability to keep compounding from there.

We felt some urgency, I will admit that, and that is still true. AI is moving so very quickly, and we certainly didn’t want to spend three years admiring the problems while others were moving forward and experimenting, so we dove headfirst into this. So we obviously spent an enormous amount of time and money on infrastructure, engineering, and data collection, but most importantly, we invest in time and judgment of our lawyers so that AI becomes a part of how we operate rather than just another piece of software that we can use or lose.

I think if we get this right, the payoff is that Certain tasks happen faster, but it’s not just that. The whole firm will get smarter. Our best judgment will travel faster, and we’ll deliver better products at a scale that would not have been possible before.

Irene Liu: Like you said, Kirkland wasn’t sitting on the sidelines.You wanted to dive deep, and you wanted to dive into it fast. And so what you were doing as part of that is you looked at over 2,000 AI products, and I was surprised that there are even over 2,000 legal tech products that with AI that you could even look at. And so it’s amazing how much diligence that you did in trying to invest in the right relationship.And you ended up building proprietary tools with Palantir. And so what did you believe Kirkland needed that an off-the-shelf product could not provide, that you decided to partner with Palantir on your next phase of AI investment?

Erica Berthou: So there is a lot to unpack in your question. There are obviously excellent third-party products out there that solve certain problems well, and we are certainly not interested in rebuilding commodity capability just to say that we have rebuilt it.

It is too costly and time-consuming, and it takes too much lawyer time to do that. I mean, the question we ask ourself is: Where does Kirkland’s specific knowledge create a real advantage? And that’s for us where building makes sense. Off-the-shelf legal tools, they don’t know how Kirkland practiced law.

They don’t contain our precedent, our experience, and our senior lawyers’ judgment. So with Palantir, we could build it the way our lawyers work. The fund formation engine, which we have built in my group, for example, draw upon model documents and investor responses and track client obligations the way we do the work.

It covers life cycle of work, if that makes sense, rather than automated, isolated tasks. So we’ve built a digital representation of our legal knowledge with a human lawyer still responsible for the ultimate advice, of course.

Irene Liu: When we chatted before in preparation for this call, you talked about Palantir’s superior ontology and knowledge graph capabilities, and that was one of the draws for Kirkland to partner with Palantir. But for lawyers that don’t really understand what ontology or knowledge graph capability means, can you describe that, and what does that actually allow Kirkland to do differently?

Erica Berthou: Sure. I would be happy to. So had you asked me this question three years ago, I would not have an answer for you. And it’s still true that I am a funds lawyer, not an engineer, so I’m gonna give you the sort of lawyer, practicing lawyer explanation with respect to the ontology.

So I’ll give you an illustration. A fundraise is not a pile of documents. It’s a network of decisions and obligations for funds and their investors. And the ontology is technology that basically maps that network, that relationship, instead of treating each document as isolated text, it’s how everything knit together.

So for example, the ontology helps us connect an investor request to a relevant provision, show how similar issues were handled, and flags obligations that must be tracked. It’s an organic living system, so it gives us as lawyers relevant knowledge in the context of the life cycle of a fund. You could think of it like this: A search engine finds information, whereas a knowledge graph, the ontology shows how it is connected.

And for lawyers, it’s in those connections where the real judgment begin.

Irene Liu: So interesting that you talk about this network of knowledge, and so that it’s really important not only to capture simple documents and precedents, but you really need to capture the judgment of senior lawyers, especially the lawyers that have done complex transactions over the years.So how do you take something so nuanced as the experience and expertise of a senior lawyer and turn it into a system that lawyers can actually use? How do you turn that knowledge into this network or connections that can be graphed?

Erica Berthou: Yeah, and it’s hard to even explain this, but the way I think about this is that judgment stays human, right?

We are not trying to create an ontology and a digital representation of our knowledge in order to create this one formula or produce one automatic answer. What we’re trying to do is basically make the ingredients of good judgment more accessible. So we make accessible, like, questions that experienced lawyers would ask, patterns that we see, the issues that we flag, and then the consequences that we consider.

So to get this done, we had our senior-most lawyers, including myself, work directly with the engineers at Palantir to explain why a provision matters, when a solution fits, and what facts and circumstances would change the answer. I wish it was as simple as handing over a document library to a Palantir engineer and say, “Hey, please build our judgment.”

But obviously that would be a fool’s errand. You basically have to unpack the thinking behind the documents, and that takes a lot of time. So the system that we have built can surface relevant experience at the right moment. At the end of the day, though, we as lawyers will be the ones understanding the clients and evaluating the circumstances and making the call.

So that’s a long-winded way of saying basically this system, the ontology and what we’re building are giving more of our lawyers access to the judgment of people that have seen the movie before, so to speak.

Irene Liu: And one of the conventional views is that law firms shouldn’t build their own technology. And so a lot of law firms have leveraged a lot of these off-the-shelf tools, mainly because lawyers aren’t engineers, and internal tools can become cumbersome to build and then to maintain, and technology moves so quickly.And I know you agree with some of that critique, and yet Kirkland chose to build. And so what made you believe that Kirkland could do it successfully, and how important were Palantir’s embedded engineers, Kirkland’s scale, and the involvement of your own lawyers to make this successful?

Erica Berthou: Right. Okay, I mean, I guess I’d like to challenge part of that premise for Kirkland, at least.We’re a very large firm, obviously, but culturally we are very entrepreneurial. We are big, firm resources with startup instincts, if, if you will. And I think our organization, our lawyers and our leaders here are expected to act like owners. So if we see an opportunity to make the business or a client experience better, the instinct here is to go after it and not spend six months to try and get permissions.

So we put our senior lawyers side by side with Palantir engineers to make decisions quickly, test things, learn what work and does not work, and keep moving. And it’s really the combination that’s the point. They bring their world-class engineering, we bring scale, judgment, and expertise. Together, we are solving the problem.

I will say, though, Irene, that I had underestimated how much lawyer time it would take to do something like this. I knew that we were going to invest a lot of time, but having 30 very senior Share Partners on the firm side spend this much time with Palantir, I guess I hadn’t quite appreciated that. But it’s been incredibly rewarding and fun and learning new things.

Irene Liu: Yeah, and that’s so interesting because it’s not just the money that you’re investing, it’s 30 knowledgeable partners’ times that you’re investing to build this as well, too, which is quite fascinating. And have you also had to invest in bringing in new different types of what we call legal engineers, essentially, people that can help you build as well on the Kirkland side to partner with Palantir?

Erica Berthou: Yeah. I mean, of course. We have a very large innovation team here, and we are needed to get together with them through every step of the process. So they give us education, strategic thinking. We have legal engineers here, and with Palantir, we are co-hiring engineers with Kirkland because at the end of the day, this is not a one-time, short period build.

This is going to be an ever-evolving product that we’re doing, so we will have very robust engineering capabilities internally. And this is not just in my group, but this is for the firm in totality. So it’s a lot of non-lawyers also involved on our side, and they learn our personalities and what works, and we as lawyers can be challenging at times.

We question a lot of things, and they know how to navigate us, so this is good.

Irene Liu: So, Erica, one of the other things that was so fascinating when the announcement was made was, wow, the law firm is willing to invest not only such a large amount, but the fact that the leadership was so supportive and that the entire firm lawyers were supportive of this investment.And so how much did Kirkland’s leadership structure affect your ability to move quickly? I know you mentioned that Kirkland is very entrepreneurial, but when people from the outside often think about law firms, that’s not necessarily the first thing that they think about, but maybe Kirkland is different.

So is there something with the leadership structure that makes Kirkland uniquely different in being able to move quickly on such a large investment and bringing everyone on board?

Erica Berthou: Yeah, I mean, it’s such a good question, and I think the answer to it is both cultural and structural. So I think culturally we are very comfortable moving ahead on opportunities before we have every answer, and you really can’t have every answer in a fast-paced AI evolution, right?

We also have a very non-hierarchical leadership structure if you compare it to a lot of other firms. We don’t have a lot of committees. We have a pretty flat decision-making structure, and that makes it easier to take decisions quickly. I think none of this signifies carelessness. It’s just that we are distinguishing risks that must be controlled from uncertainty that we manage through iteration.

And I think our lawyers are trained to approach work as owners responsible for a client’s business and the next generation of lawyers, and that really created urgency and accountability At the end of the day, leadership mattered a lot because AI cuts across every aspect of our firm, whether that is talent, data security, client service, pricing, and operations.

I think for AI adoption and development to work, it needs to be a clear operating priority and not just a technological side project, if you will, and that is what our leadership is doing.

Irene Liu: Was the entire firm supportive when the leadership came up with this decision? Was it a excitement that reverberated, or what was the response within the law firm as well?

Erica Berthou: Yeah, look, I think everybody was super excited. Like, if I were to generalize the leadership and the firm investment into AI, everybody was very excited about that. The sort of daily, “Hey, now I have to figure all of this out” dynamic, I think that’s harder for people, right? Like, think about what you knew about AI three years ago, two years ago, one year ago, like all the skills you have now compared to then.

You’re learning new things all the time on top of a very heavy workload, so I’m sure there were some hesitant people as well, you know? But all in all, I think this has been remarkably well-received, and a huge amount of people have been asking to be involved along the way at every level of our law firm.

Irene Liu: What about the clients? Were they excited? What did they think? What was the response you were hearing from them?

Erica Berthou: Yeah, I mean, I think clients are Supportive, excited, grateful. I think they like the idea of faster service at a attractive price point. So yeah, I think the reception has been remarkable.

Irene Liu: And going back to the culture aspect, you built these tools internally with 30 senior Shear partners, which is a lot of time that they had to invest-

Erica Berthou: Yeah

Irene Liu: and with associates, and you also brought in junior lawyers as well. But there’s something counterintuitive that you said about senior lawyers, that sometimes they adopted the technology faster than junior associates, which I thought was quite surprising. And you said one of the reasons that you observed is that it‘s because junior lawyers are so busy just trying to learn how to practice the law.

And so what did this transformation and AI investment teach you about change management?

Erica Berthou: Yeah. I mean, I think if we put ourselves into the junior lawyer’s shoes, I remember starting as a young lawyer. This was last century, by the way. You’re just figuring out how to go to work. What are you wearing? You have a little bit of performance anxiety.

You’re just learning how to be an employee, right? So now we’re asking them to move really fast, and they have to learn law. They have to understand the clients. They have to understand the business environment they live in and firm standards. Oh, and time management and time sheets, and the list is never-ending.

So yeah, they have so much on their plate, and now we’re developing proprietary AI tools that they have to learn in addition to the stuff they already do in their private life. So it’s a lot for them. Change management is a real thing, Irene. I mean, we worked really, really hard to make sure that AI doesn’t just become yet another thing they have to layer on top of a heavy workload.

So, you know, one of the biggest points for them is that the tools that we have built in AI can really take some of the repetitive work off their plate. There can be some soul-sucking work, I think, when you’re a young lawyer that is, like, exceedingly repetitive- Oh, yeah,

Irene Liu: definitely.

Erica Berthou: Right? And this really, I think, will make their work and life so much more happy in some sense at the law firm, and it will accelerate their development into, you know, really good advisors.

So I think once you explain that, they feel excited about it. One of the things that we did was to invite the young lawyers to give feedback on the products that we develop, and that feedback is super powerful, right? So we have taken their knowledge on board, and we give them opportunity to present. So our funds group is actually 700 lawyers, and we have regular town halls where we talk about what’s next steps, what can be improved with our products, and things like that.

And we have the younger lawyers, like really junior lawyers, present at these meetings about technology and how they think it should work, and that’s, like, a huge opportunity. It’s not necessarily the case that they would be the one speaking on the most complex le- legal issue, but they can know just as well as we can, right?

Yeah. So we try and create empowerment for the young lawyers as part of this process, and that’s been helpful for adoption.

Irene Liu: Yeah, one of the things that you mentioned is that junior lawyers’ work sometimes involves drudgery, and that’s somewhat the reality of it. And so what do you think will change or disappear for these junior associates as a result of AI, and what will incoming associates need to learn instead?

And that’s an area of concern that even law schools are thinking through as well, too, and so what’s your perspective on that?

Erica Berthou: Yeah. Look, we are not expecting lawyers that come into our firm to be completely fluent in AI legal tools, obviously. We will give them all the training and all the support that they need.

But the reality is that a lot of the tasks that they did as young lawyers will become less relevant over time. So basic data extraction, you know, basic initial document review, I think that will disappear, and they will get closer to judgment, strategy, client interaction. So I think skills that they need to hone is really Their presence, being comfortable delivering advice, having great energy, connecting with people.

These are talents that have always been relevant, but I think you could get pretty far in the legal industry as a young lawyer by being good at processing documentation, and I think that will take on a more diminished role. And generally, if you think about the work we do across the whole spectrum of law, I don’t think of myself as just a lawyer, right?

Like, we work closely with clients on business issues, strategic judgment, institutional risk. So we become trusted advisors to the C-suite of executives at clients and I think young lawyers should envision themselves as that as well and hone those skills. Yeah. They’re gonna be… You know- Yeah … like every generation is better than the next, they’re gonna be so much more impressive than we are.

Irene Liu: Yeah, and it’s really interesting because AI actually speeds up the development that they have, because a lot of the drudgery is eliminated. So like what you were saying, it’s building that trust with the client, some of the expertitis, maybe they can hone in on that even earlier in their careers. That makes them even stronger as a lawyer.

Exactly. And like you said, maybe they will be stronger, and hopefully will be stronger as partners, and hopefully maybe even earlier partners than us because they have the tools that we didn’t have earlier on in their careers- Yeah, yeah … to really be able to hone in some of the people side skills- Exactly that makes the client relationship so pivotal. Yeah.

Erica Berthou: Yeah. There is no doubt that AI will help accelerate their judgment, and I think it’s actually true that lawyers at some point in their career are starting to thinking about business development and how do you bring in a client or a matter, and they just get younger and younger when they walk into my office and wants to go to a pitch for new business, and I, I think that they will be more equipped to participate in those conversations earlier.

Irene Liu: That’s great. And one of the other things that you mentioned earlier was that the client response was of excitement, but I’m sure some of them were wondering, “What does this mean for my bills?” Does it mean that AI can make legal work faster? So at the same time, does that mean there’s less fee pressure for us, and will AI accelerate the project-based type fee structure as a result?

So are there questions like that that’s happening, and do you think the billable hour remains durable in this age of AI?

Erica Berthou: Yeah. I think every single client will ask about legal fees, because reality is that to deliver a good legal product, it’s not just about the content you deliver, it’s also at the price point you deliver it, and it relates to value, right?

You pay for value. And I think that the billable hour for sure will be muted over time. Nobody would be happier than us as lawyers if we didn’t have to record for six-minute increments or whatever it is that a client is requesting, right? And I think that there is going to be a definite move towards project-based or value-based billing, where for the right- Project, I think efficiency and pricing certainty would reinforce each other, and I think everybody loves to, at the end of the day, have a good reception, not just on the legal advice, but also on the legal fees.

I don’t wanna over predict, you know what I’m saying? Because we’re still in the early days of AI adoption, and it’s evolving quickly, and there are some clients that actually prefer to stay with a billable hour even when we have offered to do a more kind of value-based solution for them. You can pack more into an hour with AI potentially than before, and that may sp- speak for a billable hour.

So different clients will have different preferences, and here at Kirkland, we’re gonna stay nimble and are super open to having a conversation about what the right pricing structure is

Irene Liu: Are you having more pricing conversations than before because there’s so much client pressure as well as a result of AI on the fees?

Erica Berthou: I don’t think I have more. I think I have better conversations, right? Mm. Because I think that there are areas in law that have become filled with legal administration work that doesn’t feel as valuable to clients, and I already mentioned this, but they pay for value, and they wanna pay less for things that are repetitive or legal administration work.

AI solves some of that price pressure, so the conversations are better. I feel like we are more and more on the same page.

Irene Liu: One of the other things that is really interesting is that the Kirkland has the scale to make the significant impact, but what about firms that just don’t have that type of scale to be able to make that kind of investment? Does AI widen the gap between firms as a result, and potentially even the sophisticated capabilities such that the availability may vary depending on the firm that you go to?

Erica Berthou: I think that AI may widen the gap between law firms. I do think that, and I think that is just, in some sense, to our advantage. At the same time, I think AI can make more sophisticated solutions broadly available, right?

Not everything, but I think that it may give some law firms that otherwise wouldn’t have access to that solution access. But generalized models don’t produce the outcome that we are looking for, so I think the differentiators are data, market knowledge, integration or workflow, user adoption, and senior judgment.

Do I think that every law firm should try and build a proprietary product? Absolutely not. Clearly not every farm can or can afford it, but I also think, as we have experienced, really require a lot of lawyer time, so it’s really only for law firms of scale, I think. But, you know, Irene, we have a turnaround mindset here at Kirkland.

The largest investment doesn’t always win, and you have to stay humble and hungry about that, so we are

Irene Liu: And it’s interesting because while Kirkland has the scale and the ability to invest, you also see these nimble AI native law firms that are being built with a very small group of lawyers, where they are doing and trying to emulate what you’re doing at Kirkland, but at a s- smaller scale, but they’re leveraging their expertise that they’re bringing in, especially because there are a number of law firms that are being created with the expertise of very senior lawyers as well.

They’re hiring in senior lawyers from large law firms to build this ontology as well, too. And so it’ll be interesting to see how the AI landscape impacts the law firms, especially of all different sizes, because it is helping to escalate and up all of their games, if anything.

Erica Berthou: Yeah, I applaud their efforts.It’s super interesting to see them do this, and we have some retired partners that are obviously engaged in this, and I’m definitely following them and hope they’re successful. The reality though, Irene, is that what they are doing, we are already doing. We’re just doing at a scale with more knowledge, with more lawyers, and I think more capabilities at hand.

So I think they will serve a different segment of the market, and I’m not really concerned about that. We’re doing the same, but, but more, if you will.

Irene Liu: And I love how you said that you’re always staying hungry and humble Yeah … which is really cool because this is a five-year investment that you have at least initially committed to.Do you see this as being an ongoing commitment? And if we were to chat after five years, since this is year one of that commitment, what would need to be true for you to say that Kirkland’s AI strategy really worked?

Erica Berthou: Well, for me, I think AI would be just a natural part of how we work, right? It wouldn’t feel like we’re at this early innings adoption stage.

We have reached some kind of steady state, and evolution and development of AI capabilities will naturally be ongoing. I don’t think that ever stops, but it will be part of our everyday life. I think from a client perspective, we would define their success as getting access to our legal advice faster at a really good price point for both parties.

For our lawyers, I think that we know more. We are a better version of ourselves. We have even more fun when we practice. And I think when you look at the suite of products we offer as a law firm, that has also evolved. We will be in markets we haven‘t been before. We will sell legal services and products we haven’t done before, and we’re keeping a super open mind about how we scale.

I don’t think our business has shrunk in any way. We are in a total expansion mode here at Kirkland, and this really gives us an opportunity to expand our business, so success would also mean that we have evolved that way. To finish off, I think failure is not an option. We’re here to win, and I think we’re well on our way to do that.

Irene Liu: Well, I can’t wait to talk to you hopefully in 2031- Yeah … to hear where you are. Wait, how old will I be then? Well, you’ll still look young and youthful. And so I can’t wait to hopefully chat with you then, and I hope you’d be open to it then for us to pick up on this conversation and reflect on how that five-year investment has gone.

Erica Berthou: Yeah. We will definitely speak again in five years to see where we are. It’s gonna be amazing. This will be like a time capsule that we open in five years, Irene, together.

Irene Liu: Well, thank you again, Erica, for joining us today. I really appreciate it.

Erica Berthou: Thank you. It is my pleasure.

Irene Liu: A huge thank you to Erica Burzu for joining us and giving us an inside look at how Kirkland is approaching AI.Perhaps the biggest takeaway is that this is not really a five-year technology project for Kirkland. As Erica put it, the goal is for AI to become a part of the fabric of the firm, woven into how lawyers work, how clients are served, and how the business evolves over time. And that framing feels especially important because AI is already deeply connected to how firms think about talent, client service, pricing, and competitive strategy.

The real question is how those pieces evolve together as AI becomes more embedded in the way legal work is delivered. And those are exactly the kinds of questions we’ll continue exploring here on The AI Sidebar. So thank you for tuning in. New episodes drop every two weeks. If you enjoyed today’s conversation, please follow the podcast and share this episode with someone who is thinking about the future of law and AI.

Until next time, stay curious and keep learning.