Transcript of Beyond Billable - Driving Conversation with Steven ter Horst

Pim: Well, we’re rolling. Welcome to Driving Conversations. Really great that we can do this together. Let’s first start with you telling us who you are and what you do.

Steven: Yes, of course. Yes, Steven de Horst, I work at the law firm Houthoff in Amsterdam at the Zuidas. I’ve been there for a little over ten years now.

Before that, I spent about seven years at another top law firm in Amsterdam, De Brauw, and I actually started my career as in-house counsel at a bank. That was in 2007, I believe, and I did that for about two years before making the switch to private practice. Funny enough, when I joined, that was right when the financial crisis started to hit.

And I was still in the final phase of my master’s program back then. Really just trying to figure out, do I want to go into private practice? Do I want to stay as in-house counsel? Do I want something else? And then something else did happen, because the financial crisis, that was actually the starting point of all sorts of developments, innovations, changes in legal services, which I had not seen coming at all. But what happened then, and what I’m still working on today, that’s where I really found my niche. That’s where legal content, process management, project management came together in a very nice way.

And that eventually, in a few steps, led to the role I now hold.

Pim: Fantastic. And now you are at Houthoff. And of course, I’ve seen you from the sidelines a bit. And that is the reason I also invited you. Because with Houthoff you have taken a first mover advantage when it comes to the adoption of generative AI.

Steven: Yes, I think you could put it that way. We are quite far ahead. Yes, that’s a conscious choice, although being ahead in itself is not really the ambition, but we simply see this as a very important development. And with this development specifically, we very deliberately considered how we want to deal with it.

Do we want to really take a step forward in development, do we want to be on top of it and basically sit in the driver’s seat of everything we see happening now? Or do we take more of a layback position where you calmly wait and see how the market develops, and then you simply step in later in peace and comfort, maybe when things are already more crystallized? And that is ultimately also a choice you can very consciously make as an organization, and we do that for each development.

A few years ago blockchain was very hip and happening. That was also such a movement where everyone thought, this is going to have a major impact on all facets of service delivery and across many sectors. And back then we consciously said, let’s just wait a little bit and see what happens. Also because it wasn’t entirely clear to us what the relevance would be for legal services. Maybe for other types of processes, but not so much for us. And I think that turned out well in hindsight, maybe even with some wisdom after the fact. But just to sketch that we don’t necessarily want to be at the front of everything.

Steven: But with AI we did make that choice very consciously, because we very quickly saw that this is such a major development, almost a kind of generational shift that will take place. If we want to maintain our position as a top firm in the market, then by definition in our core services you’re already in an absolute leading position. If you want to secure that, or maybe even expand it, then you simply have to go all in on this development. And that is exactly what we did.

Pim: I find it interesting how you made that choice, because afterwards it’s always very easy to say that generative AI is disruptive. But at the moment you made that choice, there were also a lot of flaws being made. Hallucinations were very common, and yet at that moment you already saw that this was going to change the market.

Steven: Yes, but still we also took our time. This was of course something that, I think, kind of exploded just before the Christmas period in 2020. That immediately caught everyone’s attention.

It was like a bomb that exploded on LinkedIn. So in that period I, as a tech-interested adventurer, of course tried out everything I could find. And I think everyone, including myself, had that experience of wow, this really is a different kind of beast, so to say.

Then in the following year, 2023, we actually took quite some time and calm to investigate what it really is, what possibilities it will bring, and also what inherent flaws and risks come with it.

Steven: And we actually used that first calendar year quite well for that. Partly also to do some internal information sharing. To inform the business: What is this that’s coming our way? What could we do with it later on?

What we did already during 2023, because we saw that people were starting to use this through all kinds of channels—which of course potentially brings risk when it comes to client-sensitive information—is that we quite quickly made Copilot available, which at that time was still Bing Chat Enterprise, the web variant. That way everyone at least had a safe AI option to use.

That step was fairly easy, since it just fell within our Microsoft license. But it gave some comfort that there was a go-to tool people in the business could use.

By the end of 2023, the first real parties started to emerge with legally specific solutions in the market.

Harvey was a name that spread quickly. Of course we had heard and read about the collaboration they had very early on with Allen & Overy, and Harvey again connected on the back end with OpenAI. Well, that caught our attention.

So at the start of 2024 we did a proof of concept. At that point, honestly, there wasn’t that much choice in the market yet. CoCounsel was just a bit upcoming but still very focused on the US.

Steven: It was only a bit later in Q1 2024 that Legora came onto the market, I think. But that was pretty much it. So at that point we were really only working with one party, and that was Harvey. At that stage, Harvey wasn’t much more than just some buzz in the market. It was literally just a website with a black page:

Harvey, AI for elite law firms.

Well, that obviously grabs your attention if that’s the area you want to operate in. And a button for the waiting list. That was all. But somehow everyone was talking about it. Maybe it was the name, maybe the collaboration with A&O, but it really became the talk of the town.

That also reminded us of what had happened earlier with Ross Intelligence. I don’t know if you remember that company?

Pim: Yes, definitely.

Steven: A kind of previous wave, that must have been around 2015, 2016. As far as I know, they’re defunct now. That was probably just a bit too early. But still, Ross, Harvey, familiar characters from Suits, so the marketing worked.

And I think Harvey just handled that smartly, while outwardly they probably didn’t even put that much effort into marketing.

Anyway, of course we and other firms signed up for the waiting list. I’m in all kinds of international networks where you talk with other law firms, and everyone was in the same boat.

Are you on the waiting list? Have you heard anything yet? No, we haven’t heard anything either. So that created a kind of anticipation, like, when will we get that call? We ended up being called quite quickly. My impression was that Harvey had some sort of hit list of the top 100, top 200 firms worldwide.

If you isolate all the big firms and then look just one segment below that, then as a sizable firm you quickly fall into that cluster. And in that first conversation we immediately had a good connection with Harvey. Then we ran the pilot, and that’s when we noticed: this is something that much more speaks the language of how lawyers think and talk.

That for us was the difference compared to the generic models and tools that were available. Then of course the question was, yes, it’s still very early days, but are we going to jump in? Well, we did, with a limited number of licenses. We started with fifty people, which allowed us to cover all practice groups and all seniority levels within the business with a few licenses and then just see what happens.

How will it be used? What can we do with it? How does the tech develop? From that point we started to gradually scale up the use of Harvey. We did that very deliberately on a small scale, not with a big bang for the entire organization.

That also had to do with costs. It is and was quite an expensive technology to roll out at scale. And it’s still very much in flux. So when we started, we didn’t know… yes, this is now a tool that happens to be here, but who knows, maybe in a few months we’ll think differently about it. Maybe it’ll be overtaken by other technology coming onto the market. We wanted to keep that deliberately open.

And scaling up with small groups also gave us the space to focus very specifically each month. Roughly ten people onboarded, with a lot of attention, mandatory training, solid guidance during the first onboarding, checking how people use it, inspiring them where they could use it more, and also being inspired ourselves by use cases people in the business discovered that we hadn’t thought of yet.

So that was basically the start of the adoption plan. The funny thing is that the buzz in the market around Harvey, the anticipation, the fact that a limited number of people could use Harvey, and that this spread like wildfire within the firm, created demand we hadn’t anticipated. And that demand was much bigger than what we could serve.

So a waiting list emerged on its own.

Pim: Internally, you mean?

Steven: Internally, indeed, yes. And the fact that there were limited licenses, that the people who had a license were very satisfied, that more people wanted one but had to wait, plus mandatory training, and in the end also a “use it or lose it” policy.

That actually became an enormous catalyst for usage and adoption. It wasn’t planned that way at all, but it turned out to be a very positive side effect. And yes, for me—I’ve now been working in legal services for about twenty years—this was the first time we were really working with Legal Tech where we had such massive traction in rolling it out.

Pim: Did anyone lose it then, because “use it or lose it”? So that means if you don’t use it, someone else gets that license? Did that happen?

Steven: That did happen. And we still do it that way. We continuously monitor it. Not with the primary goal of freeing up licenses as quickly as possible for others who want to use them. The first goal is really to stimulate usage. But if in the end it isn’t used… it could be that you just don’t have the right field for it, or not the right practice where it applies well. Then such a license will ultimately be withdrawn. And we do that across all levels. So it could be a junior associate, but just as well a partner.

Pim: How often does that happen?

Steven: Fortunately, only occasionally. I’d say about two to three percent where we really had to revoke a license. Of course the question with “use it or lose it” is, where do you set the threshold? We modestly set it at roughly ten times a month.

So that means about two or three times a week you should at least be working with this technology for us to say, that feels sufficient to keep the license. That can still vary from very complex use, like data-room type analytical work on large datasets, but it could just as well be a simple email you straighten out quickly. So there’s definitely some gradation in that. But at least it’s a basic foundation.

We really do think and believe that this is tech everyone will eventually need to use every day. So that’s why we’re also cautious not to demotivate people by taking away licenses too quickly.

Pim: Can you say something about the total number of prompts being made?

Steven: At the moment we have one hundred and eighty users. The usage per month fluctuates between fifteen and twenty thousand prompts for that group. And if you work that back, that comes out to four and a half to five times per person per day.

Pim: Yeah, those are good results.

Steven: It’s quite a bit more than I had expected at the start. Back then we really didn’t know what to expect at all. But what I thought at the time was, well, if that’s the number per week, then that would already be nice, because that would mean people are on average using it once a day. But as it turns out, it’s much more frequent than that.

Steven: And on top of that there’s also the use of other tools we have. We also use a tool for legal research from SDU, called Genial. That one is also used several times a day. Copilot we already have available for quite a lot of people. That is still being used. And outside of that, people will undoubtedly also be doing some things on their own in Perplexity or ChatGPT. So if you add it all up, business and private, it’s really staggering how often this tech is used on a day-to-day basis by a regular user.

Pim: Yes, and also interesting to hear that for legal research you then choose Genial, and for the more general work you use Harvey.

Steven: Yes.

Pim: And that is necessary?

Steven: Yes, at this moment it is. If you break it down, there are I think a few components you would want to unite in the ideal tool. First there’s the foundational layer of knowledge that’s in those LLMs by default. We don’t have to do anything for that, Harvey doesn’t either. That just comes delivered by the models of Gemini, Claude, OpenAI, etcetera. What Harvey has done, and what for now in my view still makes a quality difference, is actually that legal training layer that has been put on top. A sort of fine-tuned model they have built out of it.

Pim: And that really has an effect?

Steven: Yes it really has an effect. At this moment still. It’s true of course that all the foundational models across all domains are rapidly improving.

Pim: Okay, let me just summarize this briefly for the slightly less technical listener. What Harvey did is they took the existing models and, quote-unquote, trained them with a huge number of legal documents. That process is also called fine-tuning. And that makes the model even better at processing legal texts than one of the best models already available. That’s the summary. For a long time there was a lot of criticism about this: it’s very expensive, and did it really have any actual effect? And what I hear from you now is that it did. And if we also look at the benchmarks, for example from Vals AI. That research that came out a while ago. Then you also see that Harvey ranks pretty high, maybe even the best of them all.

So yes, it’s nice to hear that it really did have an effect, despite all the criticism that was around it.

Steven: Absolutely, yes. To what extent that will remain viable going forward, I don’t dare say. But right now, for the user, it’s really a noticeable difference.

Pim: Why do you say that?

Steven: It’s in the terminology that’s applied. The words used are much more precise when it comes to domain-specific terminology.You can also imagine the same with AI that’s fine-tuned specifically on the medical domain, or any other domain. It’s also in the structure of the output. When you write a legal document, a memo, or an opinion, it usually starts with a kind of standard structure: analysis, statement of facts, advice.

I’m sketching it at a very high level here, but those are a number of components. And that logic is very much woven into Harvey’s model.

Pim: But that isn’t sustainable, you mean? Something happens, because why would that change?

Steven: Oh no, sorry, what I meant with not sustainable is whether such a fine-tuned layer will still be effective in the future. Because it really does take huge resources to train such a fine-tuned model. Enormous amounts of high-quality legal data are needed as well. And what you see is that the foundational models are trained at such a massive scale, with so much data, across all domains simultaneously, and they’re improving. So the question is whether in a year, or two, or however many years, there will still be that much of a difference between using an out-of-the-box ChatGPT and a fine-tuned legal model.

Pim: I get it.

Steven: Yes. But anyway, that foundational layer, whether or not with a fine-tuned legal model on top of it, that’s the first layer you want. The second layer you want is all publicly available legal data, for us from the Dutch jurisdiction: legislation, case law, but also websites of authorities such as the Financial Markets Authority or the Data Protection Authority. From institutions like that you also want access to the legal data published on their websites.

The third layer you want, and that is already a bit more difficult even for Harvey, is the data from legal publishers. That sits behind paywalls and is protected by copyright. That’s the layer where you find legal books, legal journals, those kinds of publications. That’s really high-value data you want to incorporate into legal answers and references in notes. And that’s data Harvey cannot simply access. So that is an area where for now we still need a separate tool. That market is still very much developing. It also has to do with the fragmented landscape of publishers in the Netherlands. Wolters Kluwer is working on an initiative, SDU is working on an initiative. The smaller publishers are really waiting to see which party breaks through, because then they can hook onto that. They’re too small to develop something themselves.

Harvey runs into the challenge that they have to arrange this worldwide, across all jurisdictions, with X number of publishers in every jurisdiction. That is difficult. Our assessment was that the Netherlands is not in the first tier of jurisdictions where Harvey or Legora are aiming to make agreements with publishers.

So in the end, this will come. There will be connections with multiple platforms: Harvey, Legora from the Netherlands, Saga, Legasis is also active in this domain, and there are more players who will open up. But that may still take a while. And for us, legal research is such a core use case that we decided very early on that we needed to partner with a Dutch publisher that controls a block of content like that, and that can also open discussions with other publishers. Hence the collaboration with SDU.

And to round it off, the last data layer you actually want to involve in these kinds of AI tools for our field of work is our internal knowledge base.

So all the opinions, precedents, memos that have been written, clever clauses that were once drafted in an SPA or in another type of document—yes, all that knowledge you eventually want to connect to an AI solution. And for that, connections are needed between systems. For us, that’s our document management system. The Harveys and other parties need to make integrations possible, so you can do a kind of meta-search across those systems.

And that is still tricky. Because iManage, which is the system we use, is also working on its own AI. So they open the gate to upload documents from A to B, but not to use Harvey as some sort of central brain to search our entire knowledge management.

So there are still a few domains that are desirable for us, but that are uncharted territory for now. And that’s where we also see the opportunity to collaborate with parties that want to focus on those specific domains.

And whether it ultimately becomes a feature within Harvey or brand A or B manages to realize it, we are happy to think along with start-ups to develop that business case. And to a certain extent we also leverage our knowledge, and the weight we still modestly carry as a firm in the Dutch market, to push things forward where we can.

Pim: Sounds like an expensive affair, having all those platforms side by side.

Steven: Yes, well, it is. But since with a number of those platforms we actively enter into collaborations, you obviously end up with different kinds of arrangements than if you just wait until it becomes a market-ready product and then come on board as a regular customer.

So it really varies per application what kind of arrangements we’ve agreed on with those parties.

Pim: The impact on the business model. This show is of course called Beyond Billable, as a prediction in the title that the Billable Hour has had its best time. How do you see that, because this obviously involves quite some costs? How does the business model of law firms change?

Steven: Yes, you can peel that off in two ways. One angle could be, what does the law firm of the future look like? I could tell you a lot about that too, with the big strategy program we’re working on. And if I pick it up from the concrete point of business model and pricing model, billable hour…

There has of course been a huge amount of discussion about that for a long time. And I actually see it in a much longer trend, and you can trace it back, I think, to somewhere in the early 1900s or maybe even further, how developments in pricing models have gone over time.

The moment the entire market jumps on the bandwagon of “death of the billable hour,” I very consciously choose a different perspective. If only to keep asking critical questions: okay, why? What’s good about it? What’s bad about it? And the same goes for possible alternatives.

I’m not yet convinced that the end of the billable hour is really here. I think it’s a lot of marketing and hype. But there will be changes, I am convinced of that.

I believe it was in 1975 in the US, a ruling by the Supreme Court where a kind of ban was imposed on what were then contingency fees, performance-based fees. Because that basically led to a certain price-fixation and price alignment in the market, so it was not allowed. Then there was a whole movement that arose in the late twentieth century toward the billable hour. That, you could say, overshot its mark, with an enormous focus also in individual performance and compensation models on billable hours.

That also has its excesses, which you can criticize. And with the turning point of the financial crisis in 2007–2008, that led to a new wave of questioning the billable hour. Is the billable hour really still the best business model? Well, what are the downsides of that model? The first thing people say is, yes, it’s basically an incentive for working inefficiently. The longer I take to do something, the better it is for the business.

And on top of that, if you need to make all kinds of investments, how do you make them pay off? If we need to spend X amount on all these expensive Harvey licenses, and we also need a package of AI tools and other tech tools, how do you make sure those investments pay off? If those tools create a huge amount of efficiency, then how do you make that work? Because you have the upfront investment, you work more efficiently, and you also bill fewer hours. Yes, there’s a certain tension there. I understand that.

The other side of the story is that it’s still, I think, one of the most transparent pricing structures, where as a client you get a huge amount of detail. That gives you an enormous openness to discuss together: what work was done, how long did it take, at what level was the work done, and did I get proper value for money?

Other pricing models also have their pros and cons in that sense. With fixed fees, you don’t always know how that price was calculated.

The tricky part in our field is, and remains, that it’s not assembly line work. There are an enormous number of variables, which means you can sometimes end up way on the left or way on the right. And that can also change multiple times during a case. Including factors you have no control over yourself: steps taken by the other party, strategy changes, shifts in direction, issues you encounter along the way. So there are many factors that make it a very dynamic world.

And then a fixed price is not always ideal for everything, at least not across the board.

Pim: May I just… because I can already hear people objecting to this. What’s your response to people who are in favor of that change? One argument of course is: if you work very efficiently, then eventually a kind of market correction happens. The problem of non-transparency is basically solved, because you can always go to someone else who does it faster and therefore cheaper.

Steven: Yes, well, I think that’s true with every pricing model. Ultimately, it’s the market as a whole that corrects it. I do think there are more segments of service delivery now that can be better calculated in advance. What’s the bandwidth of time and effort needed. And then you’re in a position where, within certain margins, you can much better predict what the actual price will be.

Maybe not for entire cases or whole segments of cases, but more and more often for components of cases. And you’ll see that trend moving toward more fixed prices in those areas. And that’s just one model, because there are of course all kinds of variations. You can also think about performance-based fees, linking value to achieving a certain result.

That model also has disadvantages, by the way, because it could happen that the result you defined in advance changes along the way. And if the best thing for the business is to win a case, but the reputational damage is so severe that you need to settle early—just to give an example—then you also run into a certain tension.

Yes, what really is the best outcome for the client?

Pim: You could of course also make agreements about that, right? That if it changes, then the price also changes.

Steven: Of course. And I think that’s the core point. Regardless of the pricing model you agree on, it’s really about having an open dialogue and ensuring proper value for money is provided.

Pim: Yes, so it’s mostly about trust and aligning the interests with the client. And yet, as I hear it, there’s still that inherent conflict if everything is done in billable hours. Because then it’s literally the case that the longer you take, the better it is for the firm, and the worse it is for the client.

Steven: Absolutely. It’s a model with inherent flaws, and for certain components of service delivery it will create tension. Which will push parties toward other solutions, other pricing models, or maybe even entirely new services. Services that are packaged more as digital solutions, for example. Those are developments that are underway.

Pim: Yes. Like alternative legal service providers that are popping up left and right. Who might take over entire parts of the process.

Steven: Yes, that is definitely a movement you see happening. And you also see service providers partnering with AI companies. LegalZoom, for instance, has recently taken such a step. And of course that starts in a certain market segment where you quickly think, well, that doesn’t touch our deeper business for complex legal work. But that boundary is being pushed further and further. So those are indeed developments that, wherever you are in the market, you need to be alert to.

Pim: Houthoff 2030. That’s your innovation program. Can you tell us something about it?

Steven: Yes. That is basically, now that we’ve made the foundational layer of tech available, in the program to further expand and scale that. And also in domains where we’re not yet served, to have good collaborations with parties that are actively developing. We’re actually shifting toward a more fundamental question: okay, we now have the tech, we see how it’s being used, but what is the deeper impact on our business in a number of years?

And for that we set up our strategy program Houthoff 2030, where we try to create a vision of what a law firm—our law firm—will look like in a number of years.

When we started with that, it was really meant as a kind of long-term horizon. At this moment it actually feels like an extremely long-term horizon, because developments are moving so fast. Well, 2027 already tells you enough. Yes, that’s already coming very close.

Pim: For the listeners: just Google AI 2027, and then you’ll know what we’re talking about.

Steven: Yes, the developments are moving so incredibly fast that it’s already hard to look one year ahead, two years ahead, let alone all the way to 2030. Nonetheless, we’re trying to develop a conceptual vision. It won’t be one single end point, but several scenarios we’re working out. It’s really more of a thought exercise to understand in which parts of the business what kind of impact might occur.

For that we’ve defined a number of domains, or “workstreams.” I won’t cover them all, but the first workstream we tackled was an analysis of all our historical timekeeping data.

We went back to 2020 initially, but in the end we used a somewhat shorter period—2023 through the first half of 2025. All time entries, and there are an enormous number, were analyzed and allocated, based on the narrative description, to fairly generic activities.

So that means things like drafting, legal research, internal coordination, client meetings. That way we created a palette of eight core activities or skills, and every time entry was allocated to one of those. Then we divided it by seniority group: junior associates, associates, senior associates, up to and including partners.

That gave us a blueprint of how, over the past years, work on those skills was distributed across all seniority levels.

The second workstream we tackled was gathering input from the business, across all practice groups and in a number of categories of case files: what is your expectation of where AI will have an impact, so what efficiency gain do you think can be achieved for each billing group or seniority group on each skill?

So for example, with M&A lawyers, where reviewing is mainly due diligence, the estimate was: well, we think that for junior lawyers, who do most of that work, tools like Harvey could make about 40 percent of that work disappear.

If you then combine the blueprint of what’s been done over the past period with a forecast from people who understand the tech, what impact AI will have, what efficiency gains are expected, then you can start building scenarios on that basis.

Those are two workstreams. We also have a workstream where we do more ideation-style thinking about new business models. That is definitely part of it too.

New services we could design, a kind of AI-driven business. There’s also a whole workstream where we think very seriously about the entire training model of the firm. How should we approach that in the future? If much of what we now consider training work is going to disappear, how do we train people, and also, what do we need to train people for?

So in total we’ve defined about eight workstreams where, based on lots of input and discussion, we develop scenarios together. And in this way we’re creating a kind of profile: what could our firm look like in a number of years?

Whether one of those scenarios will materialize one-to-one is uncertain, probably not. But the entire thought process around it, preparing ourselves for different scenarios, having options ready depending on what happens, which actions we could or should take, what it would mean for the structure of our firm—through that, we believe we are very well prepared for the big transformation that we are strongly convinced will inevitably come. That way we can respond adequately.

Pim: That sounds very thorough. Really, compliments.

Steven: Thanks.

Pim: But I wouldn’t be Pim Betist if I didn’t also have some critique here.

Steven: Please!

Pim: Have you read Richard Susskind’s latest book? Thinking About AI?

Steven: I haven’t read it yet, no.

Pim: That’s really a tip, also for the listeners.

Pim: And what he explains in there is the difference between process thinking and output thinking. And the example he gives is a fence at the top of a cliff, and an ambulance waiting at the bottom.

And what he’s really explaining with that is that the output, namely no conflict, no legal dispute, and simply being able to do business smoothly, is more important than the solution, namely resolving a dispute, which is what most legal service providers do. So the real question is, have you also taken into account black swans and this change that at some point could also arise through technology?

Steven: That is definitely something we explicitly take into account. Part of it is also doing a very thorough closing and competitor analysis. Standard tools, of course. But also thinking carefully, what are other actors in the market where we might expect more competition.

The part about preventative law, which you are essentially mentioning, is also absolutely something we look at. Because the earlier you can address something in a product, do risk assessments in advance, that also prevents the scenario where you later need to solve a problem. Because you’re already one step earlier, at preventing the problem. So those are definitely aspects we take with us.

And hopefully it’s also well safeguarded because we very explicitly include the voice of the client in this entire process.

Steven: We organize roundtables with clients to also understand from their perspective how, from the desk of a general counsel, the work is changing. Also because the internal business itself is increasingly getting access to legal knowledge and legal tools. That gives the general counsel more time for strategic issues, and that again trickles down into the rest of the chain.

A concrete example could be that internally we, with all the timekeeping data and knowledge of the Harvey tools, are thinking a lot about that due diligence example. How are we going to do very smart due diligence with Harvey in the future? A realistic scenario could be that the client says: nice that you’re thinking about it, but that work is going to be organized very differently. It will be done by parties who, with Harvey, extract those kinds of risk assessments and contract analyses and data points, and even add a layer of advice on top of it. And you will only be brought in much later in the chain, namely to really do the negotiation part.

So in that way we want to avoid looking at our services through a tunnel, but instead take a much broader view of the competitive field. Legal, but maybe even outside legal. And certainly making sure we take the voice of the client very seriously.

Pim: What would your advice be, as the big brother of all the smaller firms in the Netherlands?

Steven: Well, for firms that are still hesitant or haven’t started yet: shake off that hesitation. This is here to stay. Even if it’s not fully there yet. Whatever tool you use, make sure you use a safe one, but as an organization, dive in wholeheartedly. Anchor the change first of all with the right tone from the top. That is very important. This really is a strategic theme, and I’d almost say for many firms also a matter of survival.

Pim: Fantastic. Thank you so much for your time.

Steven: My pleasure. Likewise.