A Law Firm Billed Him £15,000. His AI Did It in 2 Minutes.

Shamil Malachiyev: Hi everyone, and welcome to this week's episode of The AI Floor, where we uncover AI adoption across industries. My guest today is the General Counsel at OneAdvanced, and he is the contradiction, in the best possible way. Legal is generally known as the department that slows things down, right? The gatekeepers, the people who are paid to worry about the risk. But today's guest is doing the opposite. He's declared June as an AI adoption month for his own legal team. He's building agents, he's talking about MCP connectors, to the point where even his own CTO asks him: are you sure you're a lawyer? So please welcome to the studio, Xavier.

Xavier: Hello, hi Shamil. Thank you for having me. I'm very excited to be on today, and I think there'll hopefully be a lot of topics that we can cover.

Shamil Malachiyev: Yeah, I'm sure it's going to be really fun. With your amount of expertise, I think all of the listeners are going to benefit tremendously from listening to your stories and your approach to AI adoption. Even for me, it's going to be really exciting to hear about all of that. Can we start with you telling our listeners, who are listening at home or on their commute, a bit about yourself, what you do, your background, and then talk about the AI adoption month that you've introduced to your department?

Xavier: Yeah, of course. So as you said, I'm Xavier. Everybody calls me X. I'm General Counsel for OneAdvanced. We're a UK-headquartered SaaS company. We serve seven critical sectors in the UK, everything from education to health to housing to government to transport and logistics. We have about nine thousand customers, and I run the legal, risk, resilience and compliance team of about thirty people. I joined about eighteen months ago now, and I have been going on a journey of how we better leverage tech to deliver the services that we deliver to our business, with a commitment, I guess, to the business that headcount shouldn't go up as a proportion. Because ultimately, I think historically the model for legal teams has been that every lawyer should do X amount of matters, so the default was always let's throw bodies at scale, as opposed to: let's really leverage technology in order to scale with the business and to accelerate scale, because ultimately we're here to serve the business. So that's what I've been embarking on in the past eighteen months or so.

Shamil Malachiyev: Awesome. And obviously you're working at a tech company, you've worked with private equity, you've interacted with a lot of technology within the previous companies you worked with. So maybe that is helping you get your head into the technological innovation age that we've stumbled upon as humanity. And on our conversation before, you told me that you're trying to embrace the technology among everyone working in your department, and you've introduced the notion of an AI adoption month. Can you talk about the thinking behind that, and how you structured the process?

Xavier: Yeah, a hundred percent. On the tech side, and maybe I should have shared this in the introduction, I've always worked in-house for tech businesses. I qualified as a lawyer in a law firm, but even when I was doing my qualification, I spent most of my training at businesses. I spent time at Amazon, I spent time at Getty Images, MTV. So I knew that I always wanted to work for a tech business, and I've done that since I qualified, which is a few years now. And then I did a stint in private equity as well, with a firm that invested in tech companies. So from very early on I've been fascinated by the interrelation between tech, people and business, and how they all come together. I'm a bit of a geek. I try to be as techy as I can. And if it's been around you your whole career, by osmosis you tend to learn that way. So that's the tech perspective.

And what I wanted to do this month: we are super lucky at OneAdvanced that we have a lot of tools available to us, and as a team we invested in a specific tool last year, an AI tool that was built for lawyers by lawyers, which is always helpful, as opposed to a lot of these generic AI tools out there. And their technology has evolved immensely in the past three to six months, like every tech company. And we got to the point where I said to myself: okay, my team has this tool available to them. Yes, they're using it for some, I don't want to use the word basic, but some very standard AI usage. But how can we really take it to the next level? How can we leverage MCP connectors? How can we leverage the ability to build our own agents? The ability to build workflows, skills, and everything else that's available? So in order to kick-start it, I self-declared June to be our AI adoption month.

And you could say enablement, adoption, I think there are a number of ways we can call it. But what we've done is a couple of things. We started the month by having this vendor come to our offices for two days, and we ran a bit of a legathon, trying to be cool as lawyers and stealing it from IT, from our tech friends. So we did a legathon for two days, where we got most of the team in a room and said: right, what do we want to achieve? What is the problem statement that we're trying to solve for? Where are the areas where at the moment the business is relying on us a hundred percent? Where are the things that we're doing that are super manual? How can we take that off our desk? How can we build self-serve to allow the business to do what they need to do without always having to rely on us? Because sometimes we're on holiday, sometimes it's the weekend, but why can't they still get their contracts out to their customers and sign deals? Why do they need us all the time? So we really broke it down to the jobs to be done that we wanted to use tech to solve for. And fast-forward, it's been about 10 days since we had that legathon, and the team have built and written a combined number of over 15 agents and skills to solve those jobs to be done. All of them are still in testing; one of them is live, so we're getting feedback from one of our internal teams. And I can obviously give a bit more detail on some of the use cases. But it's been super, super exciting. I mean, we've done more in the past 10 days than we had in the past three months. And I think that's because the tech is there, and people now realize that it is going to be part of our day-to-day. And I think, myself as a manager, as a leader: we're not just investing in our team's skills today, but also skills they can take with them into their next jobs. And to me, that's super, super important. So I actually agreed a pact with my team, which is to say: hey, we're going to invest in this, we're going to invest in you, but you need to give the time back, to invest in yourself and in what we're doing. And everybody said that we had a deal. And so that's how we kicked off AI adoption month.

Shamil Malachiyev: Nice. And how did you go about... obviously, over the last two months, there are the two large players in legal tech, Harvey, Legora. How do you make the decision between: okay, are we going to use the largest platforms, or maybe we can build something ourselves? Because the models are there, you put the guardrails on, the models are improving. Maybe we just build the skills that solve the specific problems we have, as opposed to investing in a huge per-seat model with one of the big guys.

Xavier: Good question. I think there are a couple of things in my mind. Without using the "we are lawyers" card, we're not inherently technical. And I think to build something ourselves, we would need the business to invest time, either by borrowing some of our engineers, who could have been spending this time on building stuff for our own customers, or from our IT team, or bringing in a third party. So I think by the time you've done all of that, in my mind it was quicker to down-select from a number of external players to the one that we think worked best for us. And even that was hard, right? Because as you said, there are these two big players, but there's also a long list of ten, twenty, thirty other players that all want a share of the pie. So even that was hard. So yes, you could say maybe it would be easier for you to build it internally. If we had the skill set in the team, a hundred percent. I don't think my vibe-coding skills were enough to build something that the business could then use and run with, especially as you think about all the connectors, the integrations, getting past our security team, our IT team. I think for a business our size, that is a lot harder. But yeah, why not? Maybe some scale-up companies, if they have the ability to do that, I think that would be great. I mean, I've always said that if I had the skill set, I would definitely build a legal tool from A to Z. But I think for now I'll stick to being a customer.

Shamil Malachiyev: And can you tell us, and maybe this is going to save a lot of time for anyone listening who is within their own legal department thinking: okay, this technology is coming out, there are all these 30, 40 companies providing the services, where do I start? In your experience, you've got a lot of agents right now, 15 agents, one in production and the others in testing. What have you found where the AI doesn't really fit well within particular use cases? And which use cases are the ones that are simple and very effective for any company to start with, from the get-go?

Xavier: I think I'm going to use the words you said at the beginning, which is: my job, and our job as legal teams, is to think about risk and to protect the business from risk. So we sort of used that when we thought about: right, what are the jobs to be done that are low-risk, where we would be super comfortable with an agent doing that job on our behalf? But I think the key thing is never forgetting that there should always be a human in the loop, especially as you start. Like, for example: yes, we've got these 15 agents, and yes, we're rolling them out, or we're testing them. But I foresee that for the first three to six months of these agents being live, as part of the workflow there will be a human in the loop at some point, because obviously we want to keep iterating, testing what these tools can do and the output of those tools. And in my mind, it's no different from having a lawyer in training, a trainee lawyer, working for you. The more you invest into it, in terms of teaching it how you want things to work, what you want the output to look like, what to do, what not to do, the more you will get out of it.

I think people still think of AI in many ways as ChatGPT. That is not AI, right? And asking a tool like that to compare two documents, I don't think is the best use of AI. I think for me, it's when you really think about: what are these jobs that you as a team, as a department, have to do that are very manual in nature, very repetitive in nature, very low-risk in nature? How do you automate that? Because ultimately that's how you can scale as a team, and that's how you can get your team to focus on things that are more high-value, by stopping doing those low-value things. And I think the other challenge I found is people want to tackle everything at once, and think that AI is the solution to everything. And it's not. The AI legal tool set might be part of the solution. In some cases it might not be the solution at all. But I think because there's so much hype about it, people think: I'll just throw an AI tool at it and it'll solve all my problems. And it's like, no. I think you have to really break it down to: what are the jobs that you're trying to do? What is the output that you're trying to achieve? Is AI the best answer to this? Because it might not be. There were some things, when we did our jobs-to-be-done exercise, where AI was 100% not the answer. So I think it's really going through and breaking down the tasks, looking at how risky each one is on that sort of profile, and then starting with those, and then building up how comfortable you are with those tools. So that's what I would share. But I think everybody's experience with this is potentially different. That's definitely what we did, and so far it has worked for us. We've learned a lot along the way. We're nowhere near done learning. But yeah, it's pretty exciting.

Shamil Malachiyev: For me personally, when I evaluate various departments and the people leading those departments across all of the industries, my logic tends to the point that everybody will have to become sort of a product owner for their department, thinking about technologies as well. And a lot of people are just starting on that path: well, I'll have to understand what an MCP is, what agentic AI is, what kind of tools are out there. From somebody who's actually walked the path, who started with law and then got his head around all of these technologies: if you had somebody in front of you who has their own department and was tasked by the C-level executives, or maybe they have the passion and want to start with something, what would be the correct approach to learning AI?

Xavier: Good question. I think it's just being open to learning, for one. And also, there's a lot of material out there, so I think the other challenge is: well, what are the good materials? And I think often the answer is in front of you. Again, for those people in this situation that work in tech businesses: your engineering teams, your product teams have all of this knowledge. I got taught what an MCP is a few months ago by our chief product officer, because we were having a conversation, he kept using the acronym, and I was like: what is an MCP? And then you take that information, and then you go and do your own research and your own learning. So I think it's starting from the beginning, and not trying to learn everything all at once.

And I think the best thing that has worked for me is putting it into practice as well. Because there's the theory of what is AI, there's the theory of what is an MCP, etc. But I think until you use it, until you see it in action... if I think of all the learning that I've done around the topic, I would say eighty percent or more has been learning by doing. Because you then get to see it in action. You get to see what works, what doesn't work, trial and error. So that's what's worked for me. But there's a lot of great content out there. I think attending conferences is also a great way. I think listening to podcasts is also a great way. I think there's a lot of great information, and sometimes there's too much information. But I would start with the people around you, your teams. I think what I've realized is there are a lot of great pockets of knowledge in your own teams that you're sometimes not aware of, because people are curious by nature. You might have people on your teams that are already doing this in their pastime, already doing their own research, already testing things on their own Claude account or ChatGPT account or whatever it is. So maybe tap into that knowledge to start with, and then grow and build from there. That's what I would suggest.

Shamil Malachiyev: And in terms of practical, getting your hands dirty in the technology: what kind of set of tools would you advise them? Because there are just so many things that you can do in practice. Maybe the easiest one is to go with Claude and just ask it, let's create a skill. Or maybe there are some other things that you would advise people to try in terms of practice?

Xavier: So I'll never forget this. We have a director of knowledge management, and she did a review of our IT estate as a company. And I think this is no different from a lot of companies: the figure was shocking, but I think a lot of companies only use like twenty, thirty percent of the capabilities of their current tools. And what I mean by that is, whether people are Office 365 users or G Suite users, a lot of them already have embedded tools that are AI-native. You don't need to go and purchase a tool. Again, often the answer is right in front of you. And I know that not every business has turned on Copilot, as an example, but there is still built-in AI functionality within a lot of these tools today, which could be a great starting point to sort of play around with.

And then I think if you then determine that you want to really take the next step: what a lot of providers are doing today, or if they're not, you should ask them, is doing some proof of concepts, or doing some trial months. So you could ask them to get a trial for a couple of months where you can properly test it, and also use your data. Because I think sometimes these environments are real sort of dummy environments, and if it's not with your data, you won't get as much out of it. So I would encourage people to ask vendors to set up a full production environment that you can test for a month, two months, however long it takes. I mean, that's what we did before we made the decision to go with Wordsmith, who's our current vendor. We did, I think it was a three-month trial, to really see how the team were responding to it, what it could do for us. And if you do that, you can then test a few of them as well and compare. Because again, everybody will sell you the dream, and everybody will sell you all the amazing things that it can do. But I think until you've actually tested it, tried it, and also put it in front of your IT team to think about: well, how can I integrate it with the rest of the business tech stack? I think it's really getting down to that level of detail, but you can't really do that if you don't have access to the tool. So I think, depending on where people are at, I would say look at your current tool set that you already have access to and try to leverage those. And then if you do want to take it a step further, then a hundred percent ask these vendors for trial access.

Shamil Malachiyev: I think that's a good life hack to know. Also, if you look at what AI companies have done over the past year or two in order to drive their own share value: they went after the biggest economic value. They were like, we can cut down your labor, you can use AI instead of all of your employees. Which resulted in a kind of fascinating way of alienating people from trying all of those tools. Because obviously, if that's something that could replace me, why would I add to that? How do you go about changing people's perception of what AI is? That it is just a very good tool, instead of something that is conscious and is going to run the company in five years?

Xavier: That is a big question, because you're throwing in some ethical considerations, you're throwing in cultural considerations, economic ones. And I'm not an expert in all of these things, but I can definitely share with you my perspective. I truly do not think, and I might be wrong, history might correct me, but I really do not think that AI will displace jobs in the way that is being described in the press. I think it will at some point, you know, if an employer has to make a choice between an employee that has mastered AI skills versus someone that hasn't, you can't blame that employer if in time they pick the person that has mastered those skills, because that skill set, in my mind, will be required for every job. And I think that's where the people that will be left behind, in my mind, are people that are potentially not opening themselves to upskilling, to learning, to developing, to investing in what this will look like. I've made it very clear to my team, and our CEO shares these views, that it won't displace jobs in the way that people think it will. Will it make processes more efficient? Of course. Will it allow more with less? Of course. But that might not always result in job losses, because for the people whose jobs may be impacted, there's an opportunity there for them to reskill and to look at different jobs or to do different things. And I know that's easier said than done, but I think that's why it's a shared responsibility, right? I think there's a lot of sort of finger-pointing towards tech providers and employers, but all of us, me included, we need to look inwards as well, which is: what are we doing to stay on top of it? How are we upskilling ourselves? How are we learning in order to stay relevant?

And I read this a while ago. There was a sentence that stayed with me that said AI will not replace lawyers, but it will replace lawyers that do not use AI. And that stuck with me, because I think that's very true. And you could argue, well, it's the same as when email came out in the nineties. It's the same as when the internet came out. This is just the next big shift in technology, and I think one of the biggest ones we will see in my lifetime. So I think we all forget what the transition looked like before. And obviously now there's a lot more hype and social media and everything else, but it's no different. It's a reskilling opportunity, an opportunity for people to maybe do different things and do them in a different way. I mean, personally, I know I save between four to eight hours a week by using some of the tools that I use in order to help me do my job. That's a whole day's worth of work. So I'm now able to be more productive by working the same, and sometimes less. And same for my team, right? They're now showing signs of being able to save as much time a day, but their jobs are not at risk. If anything, they can now do other things, more things, do things differently. So there might have been some contradictions in what I've said today, but I think overall, hopefully the sentiment is there.

Shamil Malachiyev: I think the question that I have in mind hearing that is: okay, let's say that we have a 30-person team, everybody starts using AI, any future work we can take on within our capacity. We're more efficient, we can take on more work, it's fine. What happens to, for example, all these students? Say I'm a legal student, I'm in university right now. The industry has changed. What are my chances? If the junior's job, the research and everything, is being done by all of the software, what should I be thinking about to make sure that I still have a place among those companies that don't really need to hire more people, because they can now do more work with the same number of people?

Xavier: I love that you've raised this, because it's something that I'm super passionate about, which is the responsibility that I think all of us have in helping the next generation, well, in my case, the next generation of legal practitioners. And there have been a couple of things written about this, which is: if, to your point, these tools might replace a lot of the entry-level work, how do you then get the entry-level people to train and be better at what they do, so they can, I don't like the analogy, but sort of move up in experience? And I don't think there is a clear answer right now. And I think, if I were to have an ask, it's for the universities, for the law firms, for all of these groups to think about what that looks like. Because what I think will happen, and this is just my point of view, is you're going to have people that, let's keep using a lawyer as an example, have done their law exams and are now ready to work. And to your point, there will potentially be fewer jobs, but also there will be an expectation that they can come into the job and do a lot of these things, because they're going to be assisted by a tool. So then the question is: well, what skill set do they need to have in order to thrive, in order to be different?

And in my mind, it's the uniquely human skills. Because as much as these tools can do a lot of things, what they can't do is be human. So I think skills like critical thinking, skills like emotional intelligence, communication skills, which you could argue are sort of basic human skills, I think those are the ones that in the future, or even now, there'll be a bigger sort of focus on. It's definitely something that I am doing with my team, in terms of really thinking about: okay, if you can now have a contract drafted in two minutes, as opposed to a few years ago taking half a day to do it yourself, what do you do with that output? What is the critical thinking you need to apply? What is the commercial thinking you need to apply, in order to maximize the time that you have now got back because the output was handed to you in two minutes? So what does that look like? And we're really, at the moment, thinking about how we over-focus on those inherently and uniquely human skills. Because I think that's what will differentiate candidates in the market. But also, I think that will be the expectation, because again, the output will be able to be done by a tool in a couple of minutes, but you will need to be able to supervise those tools. You'll need to be able to analyze the output. So that analytical, critical thinking I think is what will be key going forward. And the only way for people to get that experience, I think, is in the doing, more than anything else, because you can't read books about critical thinking. I mean, you can, but you need to do it. So I think finding those opportunities for anyone coming into the career, where they can showcase their human skills, where they can learn on the job, I think that will be the biggest challenge.

Shamil Malachiyev: Hm. And you've mentioned that drafting a contract before used to take five hours; now it can be done in two minutes. And within my profession as well, running a company: sometimes I like to vibe code a solution, for example, to use internally. And before, I had time to think it through, to sleep on it, to come back. Now, when things are almost instantaneous, I feel constantly exhausted. If previously your leadership expected you to do two contracts a day, now they're like, well, now you can do twenty. Yes, physically you can't draft twenty contracts, but the amount of critical thinking and holding things in your head, your neural connections are not adapted to be able to take in as much information. Have you noticed any of that exhaustion or fatigue with the decisions that you constantly have to make? Because all the work happens almost instantaneously, instead of giving you time to process it all.

Xavier: Hundred percent. I think the cognitive load and the context switching have never been more intense than they are already in this day and age. But then if you add, to your point, tools that can do things in half a second, it's exhausting. I mean, even just yesterday I was on three different screens running three different agents to do three different things. And those three things were completely different: one was about employment, the other one was corporate, the other one was commercial. And at one point I had to stop myself and be like: what am I doing? Because then you have to ask yourself: are you giving your best attention to the things that you're doing? Because before, people knew that if you had to review a contract, it would take you three hours, so they'd leave you alone. Or you would even allocate that time to yourself. But now you're like: no, of course I can do this in five minutes. But then it's five minutes of this, five minutes of that, and before you know it, you've spent three hours looking at 20 things, and then you're like: can I have a break now? So yeah, I fully agree with you. The cognitive load on our bodies, on us, and the context switching is super, super intense. If anyone has any tips on how to deal with that, I would be super open. Maybe you have some tips for me, I don't know. But yeah, I'm struggling with that right now.

Shamil Malachiyev: Nothing right now from me. I'm the kind of person who spends time vibe coding until three AM, because there is just the next thing. The next progression is not three hours away, it's ten minutes away. You're like, I'll just do the next thing and the next, and then you're like: it's three AM, I need to go to sleep.

Xavier: I think all of us are getting these instant dopamine hits, because the gratification is coming so quickly. You type a command or a prompt and you get it instantly back. Whereas before, again, it used to take an hour, two hours. Now the gratification is instant, so you're constantly in that dopamine loop. So I totally get where you're coming from. But yeah, definitely something that we're all going to have to learn to manage, for sure.

Shamil Malachiyev: It almost feels like that's the new kind of Instagram or TikTok. And it's good that, for example, with Claude they have the limits. So whenever you hit your limit, you're like: okay, I have an hour. What do I do now? Maybe I should go outside.

Xavier: Well, it's a bit like parents confiscating the phones of their kids, right? You're like: what is my life now?

Shamil Malachiyev: Exactly. And let's also touch on the point that we have legal teams within companies, and we have the outside counsel as well. And outside counsel is sort of like an outsourcing firm: you come with your issues, they say, we have experts, they give you the experts to do some research, six-minute increments, all this old model that has worked for a very long time. What is your perception of how it is already changing? How are those old models of billing working for you? And what do you foresee as your ideal type of interaction between in-house legal teams and the firms?

Xavier: Yeah, what a great topic. And there are so many things that you have mentioned there that I think could be worth clicking on, but I'll take your direction, of course. I'm not the only one to think like this, and I know this is true for many people in my position who instruct external law firms on a daily basis, which is that it has to move to value-based, outcome pricing. I'll give you a small example that happened to me a few months ago. I instructed a law firm to do a piece of work on a specific question, a piece of legislation. They did their work, they came back to me and responded in an email that was probably this long. The bill came, for 15,000 pounds. So when you look at the bill, they spent 90% of that time doing legal research. And they didn't use a cheap sort of person to do it; they used one of their good lawyers to do it. So receiving this is never great, because again, especially when you see the outcome is half a page, in my head that doesn't correlate. And I know there might be a lot of people listening to this being very upset by what I'm saying, especially if they work in law firms, but just stay with me for a second.

So what I then did is I took the exact same instructions that I gave that law firm, I put it in the tool that we use, and I pressed enter. Within, let's say, two and a half minutes, it had done its work, it had done its legal analysis, done its research, and it gave me a response, I kid you not, exactly the same as the one I got from the law firm. Now, I'm not saying that the law firm used AI and overcharged me for legal research; I think they genuinely spent time doing the research. But I got the answer that I wanted within two minutes. So I challenged them, because I run a P&L, I run a business, I'm conscious of costs. And their response was interesting. It was to say: well, if you had told us that you wanted us to use an AI approach, the bill could have been different. But you didn't. But let's agree on a price that makes sense to you. And then they asked the question, which is: what is this advice worth to you? So I went back with a response, and we then had agreement on what the bill looked like. It was not fifteen thousand, I can promise you that. It was a lot less. I think it was two thousand pounds that we ended up agreeing to in the end. So that's one example.

But I think that illustrates a much bigger challenge, which is: law firms are inherently built around the six-minute increment. Everything that they do, how they recruit people, how they pay people, how they record everything. So we as in-house lawyers, and I think some part of the industry, are basically asking law firms to redesign themselves. And I don't think that's going to be an easy job, because the whole structure is built around that. And I think that's why we're seeing some of these new AI-native law firms coming up that are outcome-based, are value-based. So I think it'll be interesting to see how that plays out in practice. Because for me, I will always need to go to an external law firm, but I might go to them for different things now. I might use the tools that I have available to do 80 to 90% of what I want to do, and then go to them for their expertise on that final five, ten, fifteen percent. Because what you're paying for in the end is the expertise now, I think. Because they have a unique position where they see a hundred matters a day, so you're asking them to apply that commercial knowledge that they have to your problem. And I think, when it comes to me, that's what I want to pay for going forward. I don't want to pay for drafting. I don't want to pay for legal research. I don't want to pay for bundling PDFs together. I mean, you can literally do this at the click of a button now, in half a second. So yeah, it will be a very interesting space to watch.

And there are some in-house teams really taking that to a different level. I don't know if you followed the recent article by Tom, the general counsel at Revolut. They've just built a new panel of external law firms that will be based on sort of value, outcome-driven. Because historically you'd have a panel and you'd invite law firms, and you would use them based on fee agreements, etc. Whereas what he's doing is inviting law firms to the panel, but they will review them every three to six months. They will look at productivity, they will look at the output, they will look at the value they've given, all of these metrics that I think you would expect these days from any third party. So you're already seeing some of these things happening, but I think it will take some of the bigger players, and when I say bigger players I mean your banks and so on, who are most of the revenue of law firms, to demand this before it really trickles through. But yes, I despise the six-minute-increment bill.

Shamil Malachiyev: Interesting. And if we talk about the five, ten percent, that is still going to be interesting in terms of what you want to rely on them for. It could be you doing all of that research, coming up with that short form of text, and then giving it to them like: can you confirm that this works?

Xavier: Correct. Yeah. Is my analysis correct? Do you have any additional thoughts? Is there anything that you've seen in the market for the past three, six months that might change my analysis or my interpretation? So I think for one-off things, that works. I mean, of course, for bigger M&A transactions and stuff, you're always going to need external counsel to help run them. But then I think what all of us need to do is question and query and challenge: how are they running them? For example, I don't want to be paying half a million pounds anymore, if you are talking about big acquisitions, to do due diligence. I had the same thing last year. We acquired two companies at OneAdvanced. For one of them, a law firm wanted to charge me £50,000 to review, I think it was over a hundred contracts. And I said no. I said: I don't want you to use people to do it, use a tool. And we ended up using Wordsmith, which we use. We got the report done in 20 minutes, and then they validated the output. And to be honest, I validated the output myself, and the accuracy was well over ninety-eight percent. So in some areas it's scoring higher than humans when it comes to accuracy. So I think again, it's on us as recipients of those services to challenge the firms and say: hey, if I'm giving you a matter, how are you scoping it? How are you staffing it? How are you using AI to get to the outcome? Because I think that's when you can start having those conversations around value-based pricing.

Shamil Malachiyev: So do you think the future for, let's say, legal procurement for large companies would be something like: you have a panel of ten potential firms that you want to work with, and then you just submit tenders, like, we have this job, we perceive the value to be somewhere around five thousand pounds. What can you offer? Who is ready to take the responsibility, and how much are you ready to be paid to hold that responsibility for this particular thing?

Xavier: Yeah, I don't think that's impossible. I do think, though, that in this industry it's a lot about the human relationship as well. So a lot of us will use people that we've used before, regardless of what firm they work at, because you've got a trust element. But yeah, I have zero issue, and I'm doing this today, going to one or two law firms and saying: hey, I've got this piece of work. This is what it's worth to me and to my business. This is what I'm ready to pay. Are you able to deliver on that? And sometimes I've had a yes, sometimes I've had a no, sometimes I've had people come back and say: I can't do this, but I can do it for that amount. Because then they explain why, right? Because maybe I didn't appreciate the work that would need to go into it as well. So it has to be a two-way conversation. So yes, I think the way that we procure legal services will definitely change going forward. I mean, it has to, because this model has been around for I don't know how long, 100 years. So it has to change. It's still one of those very archaic systems. And you can see it now with some of the Big Four, you can see it with some of your consulting firms: the way that they are charging and pricing their services is changing. So yeah, I think lawyers, or law firms, might be the last ones to fully change.

Shamil Malachiyev: Hmm. It is exciting, and anxious at the same time, to see how much the industry is changing.

Xavier: Sorry, I didn't mean to interrupt, but for me it's also super exciting to see how much, I'm going to use the word hype, but how much activity there is in the legal tech market. Because we started this conversation by you saying: yeah, lawyers look after risk, sometimes a bit more conservative. But actually, when you look at a lot of the PE firms, a lot of the tech companies out there, there seems to be a disproportionate amount going into legal tech. And it's really a market that's booming, definitely in the past year, but also in the past couple of years. Which I think again is a paradox in itself, when you think about who the recipients of those services are. But yeah, I think there's a huge opportunity for in-house legal teams, for lawyers, for legal tech. I think there's a huge opportunity.

Shamil Malachiyev: Where do you feel we are on the overall adoption curve, as an industry, let's say?

Xavier: As a legal industry, or as in-house legal teams?

Shamil Malachiyev: Let's say in-house legal teams, as compared to all of the hype that is out there, and the realities. And obviously, I take it you were on the show because you are probably a bit further down the adoption curve than, I guess, most of the companies that are still trying. What is your perception of how far along most companies are?

Xavier: Yeah. I definitely think that I live in a bubble. Because I think naturally, when you are wanting to adopt AI and you're trying to push the barriers around what you can and can't do and what to do with it, you start living in this AI bubble, is what I would call it. My CEO actually shared with me a very interesting article about this at the moment, which is: there is this AI legal tech bubble, and some of us live in it and some of us don't. So a lot of the people that I tend to speak to are on the adoption curve, I guess, a bit more than others. But it's not lost on me. Like many of us, I attend conferences, and yes, you do get to speak to a lot of people. And so I think there are definitely two or three camps. I think there are people that are still wary of it, maybe don't understand it as much, and are sort of closing themselves off to it, waiting to see what others are going to do before they make a decision. So I think there's definitely that camp, and I've interacted with a lot of people in that camp. I think there's another camp of: let's throw everything at it, it will solve everything. Which I don't think I'm in. And then I think there's an in-the-middle camp, which is: yeah, it will solve some things, it will not solve everything. Let's really think about what it can help us with, how it will help us. A sort of more considered approach. Because I don't think it's all or nothing. I think it has to be something in the middle. And then I think it's about how people get on that journey. And for all the stuff that we discussed before: I think people just need to break it down, start small, play with it, learn about it, test it, and hopefully that can get them adopting.

Shamil Malachiyev: And from the first season we had a lot of entrepreneurs and startup founders who were building solutions for various industries. For all of them who are thinking about the legal tech area, and they're evaluating whether to create solutions for internal legal ops or for outside counsel firms: what are the problems where you personally would be like, guys, build that, and if you build that, come talk to me in the future? What can you tell all of those starting entrepreneurs and founders who are there to help build the future that you're envisioning?

Xavier: This is like my Christmas list. I love it. I think the one thing that I've never seen done very well, and maybe not done at all, is the ability for any legal tool to truly show to the business the value that the legal team is bringing to the business. So let me explain what I mean. Legal is often seen as a cost center. And often the data coming out of legal teams is: how many contracts have you reviewed? How much headcount do you have? How much have you spent with external legal counsel? That doesn't tell the story, right? Because if you think about how legal teams are being used today, as true business partners, business advisors, there are a lot of other metrics that could and should be used to tell that story.

And I think for me the most frustrating part is when I show up to one of our leadership meetings or our quarterly business reviews. You have your chief commercial officer, your CTO, your CPO. They come in with data packs this big, because they can measure everything their team does: throughput, productivity. Now even in our sales team, we have recordings of the calls that our salespeople have with their customers, so we can analyze all kinds of metrics. And then I show up with: yeah, this is my headcount, this is the number of matters I've worked on, this is the amount of spend on external legal. It doesn't stack up, right? Because I might not be taken as seriously. I want to be able to say: well, hey, based on everything that we do, this is how we've helped drive the launch of this new product. This is how we've helped grow revenue. This is how we've helped save costs. This is how we've helped do X, Y and Z. Because there isn't a tool that can do that. Some tools can help you report, yeah, how quickly you've helped close a contract. Great. But what does that mean? What does that mean for the business? If the business had a slow year, and legal is celebrating that we've closed more contracts than ever, well, great. But if your contract value has dropped by 30% and your business has not hit its numbers, so what if you've closed 10,000 contracts more than last year? Who cares? So I think it's something that helps connect the current legal tools to the rest of the company tech stack in some ways, but also that is able to tell that story and really put a number on value and ROI, measuring and quantifying that. I think if someone was able to build that, I'd definitely be a client.

Shamil Malachiyev: Ooh, that's a good idea. For anyone listening, make sure to take it as a challenge for your own startup, department or innovation lab. I also wanted to start a new tradition for the show, which I'm kind of borrowing from other shows: what is the one question that you think would be good to ask the next guest?

Xavier: Ooh. My initial instinct is to ask more questions, because I'm a lawyer. Who is the guest? What is the topic about? I mean, obviously I know what the topic of the podcast is, but maybe something about themselves. When we talk about those things that are uniquely human, because again, I think ultimately business is about people, I think getting people to share a bit more about them, about their story, their experience, what personal challenges they face, ultimately hopefully humanizes all of us. Because we're all humans, we're all trying to strive for the same thing. It just happens that some of us have different jobs and different titles. So yeah, I think maybe a more human question.

Shamil Malachiyev: I really like that, because that's precisely what I try to bring out the most in the podcast: the real stories. Because we all live in all of these LinkedIn hypes and celebrations all the time, but not many people talk about the real stories, and that's the place where we connect. Because across all of the episodes, everyone says the same thing: the closer you get to the edge of the technology adoption advancements, the more you feel like you're behind. And when you start talking to other people, everybody feels like they're behind. You talk to people at Anthropic, and they still tell you: dude, we feel like we're behind, we should be moving faster. And it can't be true that we're all behind, right?

Xavier: No, I completely agree. And it also goes back to what I was saying earlier: that's a huge weight to have to carry, to always think that you're not good enough, that you're behind. I mean, I'm being super hypocritical here, because I always tell my team I'd like us to be two, three months faster than what we're doing now. But yeah, I think we put a lot of pressure on ourselves, and I think sometimes we take all of it a bit too seriously, and we forget what it's really about.

Shamil Malachiyev: Yeah, and I agree with that. And for anyone listening: if you see Xavier's level, that is something to aspire to. That is already a really cool level of understanding, implementing, and being able to envision those technologies within your workflows. So don't expect that you need to be, I don't know, at the level of the CTO of Anthropic in understanding orchestrations between agents. There is time for everything. Adoption takes time, and we're on the journey. Xavier, thank you so much for joining me today for the episode. This has been a fantastic conversation. I've truly enjoyed it, I've learned a lot myself, and I think our listeners are going to take a lot of interesting insights for themselves from the conversation.

Xavier: I really appreciate it, Shamil. Thank you so much for reaching out, and thank you so much for having me on. It's always nice to speak to people that are passionate about the same topics and have a point of view as well. So yeah, thank you so much.

Shamil Malachiyev: Absolute pleasure, man. That's a pod.