EP111: TAG x Garfield AI w/ Philip Young

This is our first episode we are recording in 2026, so we would like to wish all our listeners a great start into the new year. And yeah, this year is starting like the last ended pretty much. So AI is advancing still extremely fast, and every week there are hundreds of announcements. LinkedIn is flooded with updates and developments, so plenty to keep up sort of with for everyone. And Yeah, so the hype is continuing, so it's not over. So although lots of people say there's a big bubble and it's about to burst, but I'm not so convinced. I think we'll definitely continue much longer. And, um, yeah, Arno and myself here on the podcast, we are huge advocates for practical AI, so real use cases, working with customers for, for real outcomes. And we're also keen on learning how companies are using AI and automation and what business models develop around AI. So that's very exciting to, to hear. And what we are doing today is of course one of our favorite sort of podcast formats, this interview episodes. And yeah, so it's with great pleasure to welcome co-founder and CEO of Garfield AI, the first law firm in the world regulated to provide legal services by AI. Yeah, welcome, Philip. Welcome to the podcast.

Yeah, hello, hello, hello guys, and thank you very much for the invite. I'm very pleased to be here.

Great. So yeah, so very excited. So when I found out about Garfield AI, very excited. So very, very briefly, my background is in legal as well. So I used to work a long time in link letters, not as a lawyer, but from a technical side. So yeah, we're very excited to hear, to see it. And then now I'm excited to hear your insights about Garfield AI and hear more about how the platform you're building. So yeah, to kick off, yeah, maybe you can share a bit to our listeners what you're doing at Garfield AI.

Yeah, sure. So Garfield AI is, as you've said, presently unique in the entire world. And let me talk a bit about what it does, first of all, rather than it being a law firm, because I do think one of the interesting things about AI is it is going to take professional services, and particularly law, which is my profession, into a world where we're talking about outcomes. As a service. So not SaaS anymore, and not what lawyers do, which is think about procedure and process, but think very much about outcomes. And the outcome that Garfield delivers is for businesses in England and Wales, it helps them turn unpaid invoices of up to £10,000 into cash. And it does that using an AI platform that replicates the work a good law firm would do for a business if they had an invoice that hadn't been paid, or invoices, multiple invoices hadn't been paid, and they go to the law firm and say, please, can you sue this debtor? And Garfield does that on their behalf. And the reason it's a law firm is that it's providing a regulated legal service, something you have to be regulated as a law firm to deliver. And this is a very unique thing. No one, there's nothing like this yet in the world anywhere else. And it's unique in a number of ways, some of which I've described. Another way it's unique is it's the only AI anywhere in the world at the moment that is regulated to help someone go through a litigation process. And there's nothing else like that in the world too. And just to give you a quick flavor of where we've reached, and we, so we, we were regulated last March and then our attempt to have a stealth launch was blown up by the Financial Times in May and suddenly the world discovered us and my life became a lot more hectic overnight. But in the last sort of 9 months, we've already got to the point where we've got businesses of all shapes and sizes from sole traders all the way up to really big businesses on the platform. But we've got cases that chronologically have got to the point where people have, they've either got paid when they've sent an initial letter through Garfield, and that's the majority case. So well over half of all the letters that Garfield sends out get paid, which is an excellent result, particularly as they're so reasonably priced. But for those smaller number of claims that the users had to start court proceedings with the debtors being recalcitrant, a lot of those have already got paid. And We've got to the point now where Garfield is going and collecting default judgments from the court. And so there's another world first, the first AI ever to go into a court process and get a judgment that is, of course, a court judgment. It's enforceable. So it's all very exciting and really interesting, speaking as a lawyer.

Yeah, fantastic. With just what you mentioned, and sort of obviously the other companies are doing similar things, I think, in general, in terms of business, so obviously going after unpaid invoices, etc. So where do you see sort of the What's the competitive edge then for Garfield AI, additionally to just using AI, maybe?

So, to the extent we have competitors, they are traditional law firms doing similar processes, but with humans rather than a technological solution. And there's nothing wrong in that. That's the way that we've always done things until recently, because there was no technological basis to do the same service. You know, the technology just didn't exist. The problem with any process that is very human-intensive and therefore time-intensive is that human time is expensive. And even if you use very junior people in a law firm like paralegals, which most of these firms do, it's still expensive. And the consequences of this is that this market is very inefficient because an awful lot of invoices are just uneconomic to pursue because there's no point going to a law firm and saying, I'm owed £4,000. And the law firm says, well, it'll cost you £3,000 to pay us. Businesses think, I can't be bothered for the £1,000. Whereas if you've got an alternative method of delivery that can do it for more than a tenth of the cost. So, you know, the TV program that featured us talked about us being a tenth of the cost of traditional services. I actually think that's an underestimate. I think we're far cheaper even than that. Then suddenly a lot of things that were uneconomic become economic. And suddenly businesses think, oh great, I've got this aged debtors report that I can now turn into cash. So I think our competitive advantage is as follows. First of all, cost. Secondly, interestingly, quality. Because one thing you can do with technology is you can ensure consistency. And with the current incredibly fast rate of technological progress in AI, you can also ensure quality. And as has been pointed out to me any number of times, all you're trying to do in terms of an initial challenge is to at least meet the same quality level as traditional providers. We're already past that, and we're heading towards a point where you have an almost superhuman level of quality. So, you can achieve better quality as well as better cost. And you can also do things like provide a better experience. So, human lawyers, we have to sleep. Even I have to sleep, and we can't be available 24/7. And we all have a caseload, you know, we can't be available instantly if someone just fancies asking a question. Whereas a technological solution, it can be a lot more convenient. It's available 24/7. If you want to just drop in and see where your case is, or you want to ask a question of the AI, you can do at any point you want. And one of the surprises for me over the last 9 months is the number of our users who apparently at 3 AM in the morning think this is the right time to do a bit of debt. I'm normally tucked up in bed, fast asleep in the land of nod at that time. But some people, you know, they they obviously wake up and they're full of insomnia and they think, I'm just going to go and see if my invoice is, if I've had a response to the letter yet. So there's these three advantages, cost, consistency slash quality, and also convenience.

Fantastic.

Yeah, it sounds very fascinating. I guess from an AI perspective, I could easily see this also not just for sort of debt collection, but also I think for that type of legal services in general. You know, I, I guess if you, uh, have to deal with conveyancing cases, for example, that could be a very slow-running process relying on sort of caseworkers to, to move things forward. So I can easily see the application of what you guys are doing in other areas as well. That might be something that you've considered or thought about. So I guess A question from my side is, you know, what's the magic? You know, what happens behind the scenes? Is this something you guys have developed yourself, or how did you approach this from a kind of technical perspective?

Yeah, so just first of all, on the point about applying this technology to other processes. So we have deliberately built Garfield in a very modular way, and we can therefore take what we built and apply it to other processes. And one of the things we're going to be announcing in, in this quarter, the first quarter of 2026, is a couple of new products that we are quite excited about. And actually, the sort of work that we've been doing lends itself to being so applied, because if you look at court process, it's at the level we're talking about, it's a defined series of stages. And so if you could build for one stage, you can then take all the functions and how you've designed the data schemas and everything and just move it across into another stage. Yeah, in terms of where it There is obviously presently a limit in terms of the current level of technology available. The LLMs are getting better all the time, but we all know about hallucinations and there are other challenges around ensuring consistency and quality. My technical co-founder Dan and me have been very aware of that from day one, from when we started in the beginning of 2023. And The architecture of the system is therefore intended to mitigate the risks inherent in a probabilistic model. So there's basically, the way to see Garfield, I think, is it's a hybrid system. There is a deterministic expert system that sort of sits above the probabilistic series of LLM calls and things that we're doing at that level. And the expert system is in charge. So Dan amusingly likes to refer to it as a sort of sticking a syringe in my head and sort of sucking out my brain and putting it into an expert system. And there is a little bit of truth in that because it is a sort of codified form of my years of knowledge and experience, plus the procedural rules, plus other important legal things.

Yeah.

But we therefore don't use the LLMs to do things that you, that if you tried to use them for, you would get the risk of the uncertainty and the probabilistic outputs. We use the LLMs for things more like things like inference and stuff like that, where it's incredibly strong. And we also built an absolute ton of guardrails around what the LLMs can and can't do and what they can and can't say. And we think, we think this is the right approach at the current state of technology for this sort of product. And certainly the results we're seeing tend to bear that out. It might be in the future with the progress that we're seeing that either LLMs or some sort of successor technology can take over the heavy lifting of the expert system, but we're not there yet. I have to say, I keep— so everyone else in my team is much younger than me. I'm, as you can tell from my gray and rapidly retiring hair, I'm getting on a bit now. And I've always been a big nerd and I've been into computers and programming since I was a kid back in the '80s. I keep saying to all the guys in my team who are much younger than me, I can remember the days when an expert system was described as AI. All these— I mean, you might just see it as a whole bunch of if-then-else loops, but, um, sorry, if-then-else conditionals, but, um, but, but it is also another form of AI.

Yeah, I think it's a, it's a great use case for it. I mean, we, we see that a lot, obviously LLMs is the creative side of AI. A lot of the things we do sort of leverages LLMs for, for that purpose, but I think grounding it with the right information and that's obviously all of the knowledge in your head. To create a practical use case is such a good example of the power of kind of combining that, those almost, you know, the data elements with the creative elements. Like you said, put the guardrails around it, find a practical use case for it, and then plug in automation. I've just had a look at your website and it's fairly easy. You can just upload the invoice. From there on, I guess what would happen, that gets ingested and, you know, your process will sort of spark up and take the information and run it through your model. So that's pretty great because that's kind of like the endpoint of it. So what we're seeing, Sascha and myself as well, is sort of just, it's about kind of picking the, I guess from an AI perspective, if it's data extraction, choosing the right machine learning model or either off the shelf or creating your own, or from an LLM perspective, depending on your use case, there's loads of options out there as well. But I think that the key thing that you've touched on here is just the ability to scale this out and make this available to so many people without having that constraint of a sort of a human workforce. I guess it's sort of putting stuff in the right buckets, you know, if we can automate usually 70% lights off kind of automation is a pretty, pretty good start. Even if you're sort of at the 60% with some intervention, you're doing pretty well as well. And then the remainder of the automation can sort of just still have, I guess, a kind of human-in-the-loop element. I mean, it's interesting you mention about the sort of the court case, and that's the first time I've heard about that because there's a lot of talk about AI and, you know, can we actually use AI and to represent a proper person or an entity. So where do you think, sort of in the future, you know, and I'm talking far future, not just with something like this type of civil litigation, but also criminal sort of litigation, where will this go? Are we going to have your AI software plugged into an automated AI judge? So at 3 o'clock, your person uploads the invoice, and then by 4:30, they've got their result because it's been through court already.

Yeah.

I mean, there's just one little point I want to make on what you just said that I think is interesting. And I imagine it's a point that you guys will feel very firmly about, because certainly something I also feel quite firmly about, and that is the importance of data and having proper unified schemas. And speaking as someone who was once a practicing lawyer, well, I still am a practicing lawyer, I'm just practicing law in a very different, weird way than before. But, you know, at a traditional firm, and then a partner in a firm, the challenge that they have is the data is all over the place in different formats, stored by people in different ways, and it's very difficult to then use it. And often it's not been cleaned. And if you start from a digital-first perspective, you can set up proper sort of schemas for your data. You can populate it in a way that you can control, and that opens up a world of possibilities to you. So for example, Garfield interacts with a number of external services APIs. And it's very easy to do things like, for example, Garfield goes to Companies House and checks for debtor solvency and pulls their financial statements and displays them and gives a view on them.

Yeah.

When you try, when you want to bring a claim against corporate debtor and stuff like that is really easy to do in code once you've got the data, we know how, know how you structured it, but it's been very difficult for a business that hasn't taken that approach to do that. Yes. So moving on to AI judges. The master of the, master of the rolls, Geoffrey Vos, has given a number of speeches where he's publicly said that I think he thinks that that is the future, at least to an extent, at least to some values of claims, probably smaller claims. And I could easily see based upon where we are now that we could get to that point. I could see that is possible. I think there's a bigger question, which is, is it advisable? Now, I think that there is a there is a strong case for very small cases being decided by an automated system. And in a sense, we already have that in our society because if you look at eBay's dispute system or Amazon's dispute system, they are heavily automated. I think, and this is me speaking personally as a lawyer, I would be uncomfortable with it going beyond that because I think lawyers, we often get too obsessed about process and not outcomes, which is a point I made earlier about outcomes as a service. And What we sometimes lose track of is that clients who go through a contentious process are emotional. They often have emotional goals as well as sort of logical goals. And often you find as a lawyer that you're sort of providing a sort of amateur, unpaid, not entirely competent therapist service to your clients, as well as a legal service. And that, that even applies to the sort of work I did, which was really big-ticket commercial litigation, where you'd think they're all big grown adult business people, but nope. It can still be incredibly emotional for them and incredibly important. It can be a big moment in their lives. And I think that because it's also a human process, as well as a, um, just a sort of paper and pen process, a data process, I think it's important for certain types of cases that there are human judges. Mm-hmm. Because people want the vindication. They want to be told whether they are right or wrong and why by another human. So I'm thinking particularly family. So divorce is obviously very emotional for everyone involved. So I think that, that there'll always be lawyers involved in that and there'll always be human lawyers and also be human judges. Criminal again, I mean, if your liberty is at stake or your children are going to be taken away from you or something like that, I think you would, you would feel probably cheated if the decision was not made by somebody you could go and see and speak to.

But what I think will happen is that either the technology, like we are doing, replace entire areas of work, or it will, in other areas, it will support the lawyers to be an awful lot more efficient. Yeah.

Yeah.

Which won't, which won't, and I'm not, I'm not a, a doomster. I'm, I'm a, I'm an AI optimist. I don't think this will mean the eradication of the legal profession. I think the opposite. I think it's going to be more work for lawyers, counterintuitively, because a lot of, so I mean, Sascha was at Linklaters, who's a firm I know well, and some of the very key partners are all friends of mine. And a lot of work that a lot of law firms do is stuff that's economic for them to do, but there's an awful lot of work they can't do because it's uneconomic. And so there's an— I'm aware from my own area of practice about a huge swathe of work that no law firms are doing. And suddenly all that will become economic. It will start to happen.

Yeah.

And when people bring claims, people need lawyers to defend them. So I think there will just be a change in the way that law is practiced. That's my best prediction based upon where we are now. But mind you, if you guys had asked me 4 years ago, hey, Philip, what will you be doing in 4 years' time? There is absolutely no way, no way on God's earth I'd say I'd be doing a podcast talking about my world's first AI law firm. Unbelievable.

Yeah, it's a very, very good use case for it. It is, yeah, pretty, it's pretty obvious if you think about it, but I think the secret sauce is obviously the knowledge and the process on the law side, litigation side. You know, this is not a very generic, put an invoice in, you know, do a few things and understand how to, you know, to claim that money back. I can imagine there's a lot of domain knowledge embedded everywhere in this.

There's an immense number of edge cases, and this is on the comparatively simpler side of the litigation spectrum. You can imagine there's somebody like us maybe tried to climb up the value chain of the litigation spectrum, how the, I think the complexity isn't linear then. I think it's more of an exponential curve. And I think we tried to make the product so user-friendly that it's like a swan. It looks super friendly on the outside, but underneath the legs are paddling away like crazy. And I think we wouldn't have got here now, but for the fact that Dan and I and our team are able to work so well together, because I am quite an odd lawyer in that I really, I know how to program, you know, I've learned another programming language as part of this because I was a C++ guy. And for this, I'm now a Python person because I help out on the backend from time to time. And I did try learning some TypeScript to help with React, but I've decided that I can't stand JavaScript and all of its manifestations. I don't like that sort of badly typed language, but that's enabled me to talk to the technologists on a level they understand, you know, I can take legal concepts so I can say, this is how you could to a function, for example. Or this is how you might design a class to handle that. And that saves an awful lot of time. In fact, sometimes I, you know, some of my code they'll look at and they'll say, actually, this is really helpful and they'll take it forward. And just occasionally I will point out things in their code that I think could be done differently.

That's great. So yeah, with all that technology available, you just mentioned 4 years ago, you have not even thought about it, but you did it anyway. So with all these technologies available, we get very often the question from our listeners. So how do you actually get started with all these technologies and what kind of advice would you give sort of everyone out there maybe thinking about something similar?

I think it depends on whether they are the— I think, I think when you're going to start something like this, you need the right range of skill sets. And obviously for a lawtech product, you need technologist skills, i.e., coders, and also legal domain knowledge. And it's quite rare to find all of that in one person's head. I do know some other lawyers who are former programmers who are even deeper into that world than I am, but there's not many, many of us, I think. And then, and if you can find the right level of skills, you then need to find somebody who can interact with the other skill set in a really effective way. And that's not just sort of be able to impart technical knowledge in a way that's understood without needing some sort of intermediate translation is also sort of the right vibe between the founders. You know, they need the founders to get on really well together. So what I did when I decided that I wanted to build something and I had the time on my hands, obviously, to do so, and at that point it was meant more as a hobby. It became very serious very quickly when we realized we could do what we wanted to do. But I went out and I interviewed a bunch of people to be the technical co-founder. And Dan, my co-founder, was by far and away the standout candidate, you know, who was excited by the idea. He liked the mission. He liked the access to justice part of it. He was keen to sort of get involved. He liked the idea of helping normal people and businesses. And also he and I, we instantly clicked. So I think, and I think the same would apply in reverse. I think if you're a technical person and you want to build a lawtech product, trying to do it without lawyers would be a disastrous error. It would just be—

the way of ChatGPT, isn't it?

Yeah, exactly. Yeah, yeah, yeah, yeah. You could just, you could, I could easily predict what would go wrong. And in fact, we've seen examples of what gone wrong. We've had some people in the States who've tried to build lawtech products without lawyers, and then they almost got in prison for contempt of court, which is not a risk you want to take. So you've got, you then have to go out and find the lawyer that is appropriate. And you probably want a lawyer with quite a lot of experience because they'll know all of the practical ways that things do work as opposed to the ways they appear to work when written on paper. So it's about finding the right people. I think in my world, it's noticeable that LawTech has become a community. So when I go to the conferences, I see many of the same people and the numbers of people are growing all the time. So I would suggest if you wanted to do something like this, you start going to those as well, because there are technical people there, there are lawyers there, plenty of good ideas fizzing about, and it's a good place to sort of look for the right person.

So when you started the programming or the first phases of the product, I think vibe coding wasn't there yet. So are you now using something like that as well these days?

Yeah, so there's many, well, I mean, we've changed our frameworks anyway, because the first prototype, we were thinking it was more of a mobile product. And so we built it in Flutter. Yeah, which was really interesting for me because the last time I'd done any serious program was about 15 years ago. And of course, IDEs have moved on and frameworks have moved on. And I was quite taken with Flutter, and I was particularly taken with the language that Google developed for Flutter, Dart. I thought Dart was a very elegant, efficient language, and I really liked it. But then it became obvious to us very quickly, this is really a desktop app. We do have mobile functionality, but you need a lot of screen real estate, so it's really desktop. So we changed our framework and moved away from Flutter. And in terms of the programming team, I don't think there's any programmer on the planet now who does not, in Visual Studio Code or whatever IDE they're using, whether it's Curse or something else, they do not have LLMs built in or they're using Claude Code or something like that. And those products have just moved forward leaps and bounds. I mean, when I started to program in the 1980s, so everyone starts with BASIC and it's like 10 PRINT HELLO and 20 CODE 210, isn't it? And you ruin Dixon's Sales place, because none of the salespeople know how to turn that off. But now, you know, you've got LLMs that will sit there and you can almost just tab through whole completed functions that it will extrapolate from what you're trying to code. It will just say, here you go. I think the issue with vibe coding is that, and we're all very conscious of this in Garfield, is that it helps to template and prototype. But it's not yet really at the stage where it's production ready. So you have, you can, you can quickly run up a prototype, but then you've got to go through it and then you've got to tighten it up and make it proper production code.

Yeah.

No, I guess having the domain knowledge is, and the technical knowledge, I think that, that is such a great sort of thing to have because I suppose you could kind of generate a little bit of the code if that's what you do and sort of test it and and see it. Whereas I think if you, like you say, if you have those two things a little bit more disparate, probably be very hard to develop a product like this. So I really like the approach. I think it's such an amazing story, actually. I'm curious where the name Garfield AI came from. Is this somebody's pet name or?

No, it's a bad in-joke that has stuck. So I've had a friend from the late who was at Baker McKenzie with me called Garfield Smith. And he's another lawyer. He's got a boutique firm that he's the founder of, and he specializes in, he's a technology lawyer and he specializes in helping early stage and startup tech companies with legal advice and legal work. And he's really, really good. He's one of the best people in that market by a country mile. And he always joked to me that because I'm a big nerd, I should never have been a litigator. I should have been a tech lawyer like him. And I always would joke back, you know, I'd say to him, Garfield, you're very argumentative. Usually to be a litigator. And this sort of argument went on for about 20 years on and off. And then when I started Garfield, I said to him, mate, I think I've become a tech lawyer. So I think you've won. And so I've got to call it after you, just as a joke. And it wasn't intended to sort of stick. It was intended just as a sort of project name, you know, while we were still developing it. But no one could think of a better name. All my family tried, Dan's family tried, I think Garfield tried, lots of the other people involved in the project tried, you know, we looked at every name you could think of, but Garfield, actually, everyone just likes. It's like, it sounds like an English barrister. It's got memorable. Everyone seems to remember it. And in the end, we just are just, yeah, it's Garfield.

That's great.

So Philip, as I mentioned earlier, we have a tradition on the podcast. So we and our listeners like to get to know our interview guests a bit more. So, um, so more than just sort of the technical parts and the business side. So for that, we have a couple of questions. Is it okay to Fire away.

Yeah, go for it. Yeah. Yeah. Yeah.

All right. Okay. But the first one, who is your idol and why?

My idol? Um, you know, I don't think I've got one, but there's plenty of people I think are impressive in different ways and that you can learn things from. And, uh, so for example, I have a slight tendency to annoy people by quoting to them wisdom from either Warren Buffett or his business partner, Charlie Munger. And I think a lot of what they said, and they dressed it up as sort of homespun wisdom, but what they said was basically just very sensible things. And one of the key things that they, I mean, Mr. Buffett's still about, but Mr. Munger has unfortunately passed on. But Mr. Buffett continues to say is that if you want to look at behaviors, all you have to do is look at incentives. And I think having worked in organizations that are very functional and having also had some experience with some dysfunctional organizations in my time, it was always the incentives that was the difference between the two and also the culture that the leadership generated. And I think a lot, whenever you see things that go right in the world or go wrong, I think all you have to do is look behind that at the incentives that's driving the relevant people's behavior. Let me, he, those guys are high on my list of quotable people. And yeah, I suppose there's many other people who in different ways I think are very impressive. Not sure I could think of a big list off the top of my head, but if you gave me some time, I could.

That's fine. That was lovely. Thank you.

I've got one for you as well. Imagine you can get all the contents and knowledge of a book instantly. Which book would you choose? Or if that's a really hard one, what is your kind of favorite reading material? I guess there's a little bit AI involved in there, but maybe non-AI topics as well. You know, what's the sort of thing that you like?

Let me do both. So if I hadn't done this, one of the things I might have done, assuming I would be allowed to do it because it would be a distraction from being retired. But I always really wanted to do a physics degree and it would be really helpful if you could just sort of like in The Matrix where Neo just learns jiu-jitsu instantly. It would be really helpful rather than have to spend 3 years doing the physics degree that you could just have it all sort of put it into your head. That said, I'm not going to be volunteering for one of Mr. Musk's Neuralinks in a hurry because I don't, I think that technology's rather dangerous. So yes, that would've been good. I mean, I've spent the last couple of years sort of trying to bring myself up to date in the current world of programming. And I've read lots of books and seen lots of tutorials and stuff on that. Because actually, a lot of people, a lot of things have been learned in the last sort of 20 years that are useful for me to know. And in terms of reading, I sort of read a wide range of stuff. I have this permanent battle with a giant pile of books where every time I think I've made progress into it, I then accidentally wander into Waterstones and make the pile longer again. It's sort of my curse of my life. And I read a wide range of things, biographies, history, science books, anything really that catches my interest. I think you'd have to describe me as a very curious person. I'm basically interested in pretty much everything and I go through different interests. I might be doing economics one month and then I end up doing—

Yeah.

Presently I've got a biography of Theodore Roosevelt that I'm going through. So that's like very random from the previous topic.

Amazing. Thank you.

Right. So yeah, what is the best advice you have ever received?

I think the best advice I've ever received is around integrity. And it was when I was a junior lawyer and One of my mentors, he just liked to come out with sort of useful advice. I think he just regarded that as his role. And in a sense it was, but he would always make the point that your integrity and your reputation are something that you, first of all, you have to always guard the former and secondly, build the latter. If you don't guard the former, you lose the latter instantly and you can never get it back. And so he always said, try to do things properly and in the right way at all times. And also always try and do the best work you can as though it's your last. And that sort of informed my entire legal career. You know, I always tried to sort of meet those sort of precepts. And also now here at Garfield, and one of the reasons I was so keen to get Garfield regulated, partly I had to, but partly I wanted to, was because I felt it was the right thing to do. And I felt that if anyone was going to do it, it was sort of helpful for the regulator, it'd be a group like us. As I think the If you, if you look at the world that we're now in, and you talk to the relevant stakeholders, by which I mean the, the Ministry of Justice, HM Courts Tribunal Service, senior judiciary, regulators, practitioners, the fear is not so much AI replacing lawyers.

The fear is more lots of people not having access to a solution and then just trying to ChatGPT their way through a very important court process.

And that's the great fear that all of these people have got. And And there's a sort of secondary fear that someone might try and solve that. And there'll be a bunch of tech bros from California with good intentions, but no real understanding of the world they're getting into and build things that seem to be more plausible, but are equally, ineffective is too strong a word, but equally not quite right. You know what I mean? And so I wanted to be regulated as well to show that you can have teams of lawyers and technologists who can build things. That are appropriate and proper and built to the right professional standards that are going to be regulated and are going to provide that sort of regulated ecosystem that we need to get to.

Amazing. So the next one is the final of these questions. So if you could be an Olympic athlete, what sport would you choose?

Okay. Yeah, I'm definitely not an Olympic athlete. Unless there's an Olympic sport for eating biscuits. Which are the, which are the sort of nemesis of my life at the moment, because quite a lot of my life is being spent sat down presently. I don't think I'd be, I think I'd want to pick an Olympic sport that I actually really enjoyed. So it would be something like skiing. So I really enjoy skiing, although I'm getting on a bit, so I now ski much more sedately than I once did. So I'm going to go with skiing because it's a wonderful thing to do. There's nothing like being on top of the mountains with a clear day with no clouds and just sort of pootling around. So that would be my, if anything, if I could retire and I was able to do a bit more skiing, then nothing would make me happier.

All right. So yeah, I think that's sort of the end of the questions. Thank you very much for going through those. So yeah, it was a great pleasure, Philip, to have you on the podcast. And so I'm sure some of our listeners would like to get in touch after that episode. What would be the best way to reach you?

Well, the best thing to do is to check out our website, which is www.garfield.law. If you wanted to email me, then it would be philip@projectgarfield.co.uk. Although I do get an awful lot of emails every day, so I can't necessarily promise that I will be responding particularly swiftly just because of the volume of material I get.

All right.

Okay.

Yeah, we'll add all these details, um, um, if you like, onto the show notes, and then you will get even more emails.

Happy days.

I can also put the link maybe to your LinkedIn profile if you like.

That's good too. Yeah, no, that's another way for me to get more messages.

Perfect. All right. Yeah. With that, we're coming to an end of this episode. Thank you again, Philip, for joining us today. And yeah, and also a big thank you to all our listeners. We will be back with another episode very soon. And until then, let's automate it.

Thank you. Goodbye.