On this episode of The Geek in Review, we welcome Philip Young, co-founder and CEO of Garfield AI, the first AI-powered law firm approved for practice by the UK’s Solicitors Regulation Authority (SRA). The episode kicks off with a discussion of recent stories that explore AI’s evolving role in legal proceedings, such as avatars testifying in court and the ethical challenges that arise when deepfakes and synthetic personas enter the legal process. Philip, a seasoned litigator and technologist, draws from his 25 years of legal experience to weigh in on the potential and perils of AI-driven courtrooms, emphasizing the importance of authenticity and trust in legal proceedings.
Young shares the backstory behind Garfield AI, which was inspired by a real-world problem faced by his brother-in-law, a plumber who struggled to recover small debts from non-paying clients. Seeing an opportunity to help small businesses navigate the small claims process efficiently, affordably, and with minimal friction, Philip set out to build a system that mirrors what a traditional law firm would do—without the high cost or time burden. Garfield reads invoices and contracts, verifies the legitimacy of claims, guides users through pre-action letters, claim filings, and even court preparation, all while remaining compliant with UK legal standards.
One of the most unique features of Garfield AI is its dual design: it serves both pro se claimants and can be white-labeled for use by traditional law firms. Young explains how legal professionals can integrate Garfield into their workflows, using it to generate documents under their own branding while Garfield handles the backend. This hybrid approach provides flexibility for users, whether they prefer a self-service platform or seek a human-in-the-loop experience. Garfield’s early success has sparked interest across the legal spectrum—from solo practitioners to regulatory bodies—demonstrating that AI can support, rather than displace, the legal profession.
The conversation also delves into Garfield’s journey to regulatory approval. Young describes the rigorous process of working with the SRA, ensuring the platform aligned with legal duties to clients and the courts. He highlights the importance of maintaining accountability and explains how Garfield was rolled out cautiously, with layers of human oversight and a roadmap toward data-driven, risk-based review. With increasing inquiries from international regulators and courts, Young sees the platform as a potential blueprint for improving access to justice beyond the UK, although he notes that success depends on a supportive regulatory environment, judicial openness, and sufficient technological infrastructure.
Beyond the tech, the episode emphasizes the human element of law. Young passionately advocates for AI as a tool that enhances legal practice rather than replaces it—freeing lawyers from mundane tasks and enabling them to focus on strategy, advocacy, and client care. He shares his hope that Garfield AI and similar innovations will close the access-to-justice gap by enabling small-value claims to be pursued cost-effectively and fairly. As he notes, AI may never replace the human lawyer’s emotional intelligence and presence in court, but it can certainly help more people get there.
To learn more about Garfield AI and its innovative approach to legal automation, listeners can visit www.garfield.law. This episode is a must-listen for anyone interested in the intersection of law, technology, and the future of justice. As always, the podcast ends on a warm note with music by Jerry David DeCicca, underscoring a thought-provoking conversation that blends legal tradition with the tech of tomorrow.
Listen on mobile platforms: Apple Podcasts | Spotify | YouTube
[Special Thanks to Legal Technology Hub for their sponsoring this episode.]
Blue Sky: @geeklawblog.com @marlgeb
Email: geekinreviewpodcast@gmail.com
Music: Jerry David DeCicca
Transcript
Greg Lambert (00:00)
Hi everyone, I’m Greg Lambert from The Geek in Review and I’ve got Stephanie Wilkins from Legal Technology Hub with us today and we’re going to talk about the head-to-head analysis.
Stephanie Wilkins (00:09)
Yes, hi Greg, thanks for having me. ⁓ Yeah, one of the types of content that you’ll find in our LTH Premium subscription is what we call a head-to-head analysis, and we’re in the process of working on a few more of these now coming up soon. what they essentially are are they’re market reports on a very niche area of legal tech, where there’s a relatively small subset of tools that address a very specific use case or problem. So in that head-to-head article, we’ll explore the problem first.
Then we set out why it’s a pain point for lawyers or why it’s manual or time consuming. And then we compare in detail the different providers in the area. So we’re usually talking about a space where you have, say, four products in the area. And we’ll provide a grid with details about each of those four products in terms of the company itself, when and where it was founded, who the founders are, et cetera. And then the product where we’ll list all the major features a user would expect for a product in that niche category.
And then we identify which of the products have those features or how those features are deployed within each of the products. Or if a product doesn’t handle one, for example, that gives you a chance to do that head-to-head comparison. We also touch on security, data privacy, and data hosting information for each of the products. So buyers can see the grid and assess relatively quickly which of the products is more likely to suit their individual needs. ⁓ Additionally, we of course also on the
on our database, on our directory itself, we offer a compare feature that always allows you to compare ⁓ any product against any four other products for fewer integrated format. But our head-to-head goes far deeper into the product’s features than that compare feature does. So we’re providing a more granular comparison on a very specific problem area. In the past, for example, we’ve done head-to-heads on tools that automatically create discovery responses.
structured diagramming tools and deal intelligence solutions. And we have quite a few more of these lined up for release in the coming weeks and months. I know I have a couple of ideas myself on discovery and litigation areas where there’s a small number of tools handling very discrete but important problems. But if you keep an eye out on legaltechnologyhub.com and make sure you sign up for our free newsletters, you’ll get notified when the next time the next hedge ad is released.
Greg Lambert (02:32)
Sounds great. Thanks, Stephanie.
Stephanie Wilkins (02:33)
Thanks.
Marlene Gebauer (02:41)
Welcome to The Geek in Review, the podcast focused on innovative and creative ideas in the legal industry. I’m Marlene Gabauer.
Greg Lambert (02:48)
And I’m Greg Lambert. And this week we’re speaking with the founder of what’s being called the first AI powered law firm approved for practice in the UK. Philip Young is a co-founder and CEO of Garfield AI. And just a little backstory here in April of 2025, so just a month ago, ⁓ Garfield became the first solicitors regulation authority or SRA regulated law firm.
Marlene Gebauer (02:56)
So exciting.
Greg Lambert (03:18)
in the UK that uses AI at the heart of its service delivery. The platform helps small businesses recover debts up to £10,000. For us in the US, it’s about $12,500 or so. And it uses the small claims process in England and Wales quickly, affordably, and with very little human interaction. ⁓ big introduction. Philip, thank you very much for taking time to talk us on The Geek in Review.
Philip Young (03:48)
Thank you for the introduction, very nice to meet you both.
Marlene Gebauer (03:50)
Yeah,
thank you. Thank you for joining us.
Greg Lambert (03:52)
Yeah.
But, um, so before we jumped in, uh, I was reading a story this week where a court allowed here in the U S, uh, and I don’t have all the details here, but a court allowed the family of a victim of a, I believe it was a DUI accident who had died in the accident. Oh, road rage. That’s right. It was road rage, uh, to
Marlene Gebauer (04:15)
Yeah, and always a road rage accident.
Greg Lambert (04:20)
basically recreate this person using AI to present to the court this person’s own life story. So Marlene, I know you’ve read a little bit about it as well.
Marlene Gebauer (04:34)
Yeah, I saw
that David Hobbie posted it and I was just like, my gosh, this is kind of amazing. yeah, it was, I mean, it’s a sad story, but also, you know, from a technology point, just a real fascinating story. And what they did is they created an avatar of him and they used his likeness and his voice, you know, to just basically talk about himself. And essentially, you know, he was saying, I’m sorry, we met the day, you know,
the day we did because, otherwise, you know, we might’ve been able to be friends. And from what I understand, like everybody was in tears and it just, it’s, it’s an example of the real impact that, you know, avatars and real voices, you know, can make on people.
Greg Lambert (05:20)
Yeah, yeah. So and, Philip, I don’t know, ⁓ in the UK, if you have road rage, like we have here, like here in Houston. ⁓ But it’s a real, real thing here. ⁓
Marlene Gebauer (05:31)
It’s a real problem.
Philip Young (05:33)
We
do have this in England as well. I’m not sure we’ve got quite to the point where we have AI avatars of people that were actually a victim of the incident appearing in court. That’s very much for the future. I was saying to people about two years ago when when ChatGPT 3.5 was released and then ChatGPT 4 was released that I felt a bit like a 19th century canal owner who’s standing by the side of his canal enjoying his business and then he sees George Stevenson’s rocket go past on rails and thinks that is the future.
And that very much with this sort of story and other sort of things I think it’s very much sort of might be my mindset for the last two years.
Greg Lambert (06:03)
Yeah.
Yeah.
Yeah. Well, in, this kind of sort of dovetails in, in, don’t know if you, if you saw this story earlier from a couple of weeks ago where someone in, I believe it was in New York was doing an appeal and in front of us, I think it was a seven panel, a judge panel, ⁓ and, was giving a video and it turned out the video was an AI avatar, ⁓ that, and, the
And immediate, mean, it was shut down immediately. The judge figured it out. She was not not having it, but I have to say, I felt kind of sorry for the guy because I think he was honestly trying to do something with the technology to help himself that wasn’t, I don’t think it was really that harmful. It was no worse than reading a script ⁓ to the court. ⁓ But he was saying, you know, look, I get extremely nervous when I talk and especially, you know,
Marlene Gebauer (06:42)
I said, hell no.
Philip Young (06:43)
Yeah.
Greg Lambert (07:08)
in these situations. So I thought this would be a better way for me to present the story to the court. ⁓ You know, in that case, the judge wasn’t having it, but I think it might be something we see more and more.
Marlene Gebauer (07:13)
my story.
Philip Young (07:22)
There’s also a risk here as well. So I saw a video only about a month or so back of somebody who was on a Zoom call and they had put side by side what they were actually doing on the Zoom call, which is walking around the room and standing up versus what the people who were watching the Zoom call could see, which was them sitting down, staring at the camera, being very professional. And I was a litigator for 25 years in practice in London. And I sort of broke out in a cold sweat when I saw that because I thought, goodness me, so many hearings now in the English
court system are operated with some evidence remotely and if you can’t trust what you can see then you have to then think about what safeguards and guardrails you build in to protect the authenticity of the evidence that’s given and obviously there’s going to have to be more thought applied to that because judges have got to know who’s in their court, who’s in attendance and got to able to see the witness and what they’re actually doing. ⁓
I mean, in a fraud case, it could go from trying to detect something from the witness’s demeanor, and you can’t do that because they look perfect. They look very credible, and they’re not sort sweating under the pressure suddenly. and I imagine, sorry, Marlene, you were gonna say something.
Greg Lambert (08:31)
You’re again, Marlene.
Marlene Gebauer (08:34)
I had said, it’s like, sorry, pod dog was, was, uh, doing her thing. And so I was like, I had all these good comments and nobody heard him. I was just saying, you know, as the whole deep fake issue, um, this is something that you really have to think about in terms of, of admitting evidence and, and, and Philip, what you were saying as a litigator, like if everybody looks perfect, like you can’t read, like you can’t figure out what their tells are and, and, you know, be able to take advantage of that as a litigator. So totally understand that.
Philip Young (08:40)
Yeah.
No,
Greg Lambert (09:00)
Yeah.
Philip Young (09:01)
you’re right. It does apply as well to things like settlement negotiations because if you’re in a room with the other side you want to be able to say as you say what their tells are. So there’s quite a lot of thought that’s going to have to be applied to this to work out how to protect from mischief occurring.
Greg Lambert (09:01)
Yeah, we.
Yeah, we have one specific litigator at our firm who’s like six foot eight. And when COVID hit and he was having to do all of his work remotely, he says it really took away his advantage of being an imposing character there in the courtroom. So yeah, yeah. We’re all the same height on Zoom, I guess.
Philip Young (09:30)
You can’t sort loom over people anymore is what you’re saying.
Marlene Gebauer (09:35)
Ha
That’s right. That’s right.
All right. So pod dog may or may not come in ⁓ to join us. And so I’m going to just soldier on here and hope for the best. So like Phillip, if we were interviewing you this time last year, you know, we’d, we’d be talking about you being a pretty formidable litigator at, at cook young and Keaton, a firm that you co-founded and we’re named partner. But today.
Greg Lambert (09:49)
Good luck.
Marlene Gebauer (10:06)
People of course know you from your work at Garfield and this marks a really different chapter. So what led you to create an AI powered litigation tool and how did your personal personal or professional experiences influence the path?
Philip Young (10:22)
So yes, the road to Garfield is a slightly…
uneven one so the necessary back story for me is that I’m not just a lawyer I’m also a very big nerd and like a lot of my generation you know when the 1980s happened and we were young we had the big explosion of 8-bit computers and in this country we had we had I don’t think there were computers that particularly made their way into America but we had the ZX spectrums that had the little wobbly keys and then we had the BBC micros that were that were everywhere because every school had them they were promulgated by the BBC the government was very keen on
on them and they really made an entire generation of computer literate. So I learnt to program then and I’ve always enjoyed mucking about with computers. Fast forward a bit and I was in practice for 25 years. was originally in practice at Baker & McKenzie, which is a firm you guys will have heard of of course, because you don’t get more global than Baker & McKenzie and of course founded in Chicago too. And then in practice for 13 years at one of the firm I co-founded, CYK, which is an excellent London.
litigation and arbitration boutique. And I then decided post-COVID that I’d spent rather too much time staring at people over the screen. Lovely though it is, it’s not as engaging as seeing people in real life. And if you are litigator, you tend to be quite gregarious and all of those sort of characteristics. So I thought I deserve a break. So I retired and I retired in the sense of English partnership, not in the sense of age, because I’m not quite old enough for retirement yet, although I’m heading that direction.
And then I went across America for three months on a road trip from Boston to Seattle. we went on the northern route. So we went ⁓ up into New England and then across below the lakes, Cleveland, Chicago, across the Great Plains and Wyoming and Colorado. And by the way, I know I’m mostly talking to an American audience, America is a fantastic country. Just amazing people, amazing scenery. It’s just a shame that I couldn’t spend more than three months in America.
was
there, ChatGPT 3.5 was released and I obviously had time at my hands and I was a big nerd so I had a bit of a play and the more I played with it the more I thought gosh this thing is going to really change the way that legal services are provided. And I stopped because I was having far too much fun in your country and then a few months later in sort of April May time 2023 ChatGPT 4 was released and that was a big jump in capability.
And then I thought, well, I’m not in practice. I’ve got time in my hands. I could do something that would engage me because it’s nerdy as well as legal. What do I do? And then what I thought about was my brother-in-law, a lovely guy called Andy who lives near Sheffield, a city in England. Sheffield may not be at the top of the cities that Americans think of when they think of England, but if you think of people like Sean Bean and the musician Joe Cocker, they all come from Sheffield. Very much a northern city, very much an industrial city, lovely place.
So Andy’s a plumber and Andy’s problem from time to time is a problem all small business people have which is that sometimes people just won’t pay him. Which is a problem for him because if it’s insolvency well that’s just a risk of doing business. If there’s a problem with the service well that’s something he needs to fix but often it’s just people that know that small business people do not have the time ⁓ to spend chasing unpaid debts and certainly don’t want to be filling in court forms because that sort of thing is daunting.
As a brother-in-law, would help him. A little bit of a change from city practice, but I would help him get that sort of money. Because for somebody like Andy, a sum of £2,000 to £3,000 is a summer holiday. It’s really, really important. while I was doing that, I always thought the small claims process in England and Wales, although it is designed to be simplified, efficient and quick, inevitably it’s not really any of those things. And I always thought there’s too much friction here and we need to have a better solution. And so I started to think about Andy.
when I saw ChatGPD4 and I thought why not try and build something that helps the Andes of this world. And so that’s where Garfield comes from. It’s really an attempt to make my brother-in-law’s life a little bit easier and all the people like my brother-in-law’s life a little bit easier. And if you want I can talk a bit about how we started to build the team or you’ve probably got another question.
Greg Lambert (14:44)
Yeah, no, no, let’s let’s dig in there a little bit more. ⁓ Talk to us about how you worked on the the the development of it. Any kind of iterations what it what it looked like initially versus what it what it looked like when you finally was were ready to present it. So if I were new to Garfield, how would you explain it to me?
Philip Young (15:14)
So Garfield is essentially an attempt to replicate what a good law firm would do in helping somebody go through the small claims court system in England with a debt claim. So a sum up to £10,000 as you mentioned earlier Greg. And so the way it works is very much it follows our procedural rules. So when someone comes into Garfield and they’ve onboarded and they’ve passed all the AML and KYC requirements that every law firm
in the world now has, they will drop an invoice or many invoices into Garfield and if they’ve got a contract written contract they can drop that in as well. Garfield reads those documents, it pulls out the relevant information, it checks whether or not the claim is a valid claim. So if for example it’s over the £10,000 threshold or there are issues that would mean it’s not a small claim in reality or there’s a problem of limitation or say there’s an arbitration clause in the contract or all sorts of invalidating features then Garfield can’t help you.
If you are a genuine EAs small debt claim, then Garfield can help you and then we’ll tell you what it’s understood and then we’ll recommend the next step, which in England is a pre-action letter that sets out the demand for payment of the debt. And we have a protocol in England that says what the letter has to say. if you read it, you’d be like, that’s all common sense. It’s all, how much is it there? What’s it been incurred for? When’s it due? What the interest rate is? And then Garfield takes you through all the steps of the process. So if you don’t get payment,
in response to the letter Garfield will draft the claim form and the predictors of claim which are the originating documents that the court here needs to start court process and Garfield will then once you’ve approved those send those to court the court issues and serves them and then Garfield takes you through whichever way your case goes so if the debtor pays you which we hope of course that they do then Garfield is finished Garfield has done its job if the debtor says I will pay you but I need time and indulgence Garfield will handle that
If the debtor ignores the case entirely, Garfield will help you apply for a default judgment and send those documents to court. And finally, in the relatively small percentage of cases every year that get defended, and statistically it’s about 2 to 5%, Garfield will help you go through the steps of dealing with the defence, doing document production, which is a very limited thing in the English Small Claims Court, and then take you to the very doorstep of the one hour Small Claims Court trial. for the trial stage, Garfield generates a little PDF court bundle of
documents, a little skeleton argument, which you guys would call a brief. I always wonder why they’re called briefs, because whenever I’ve seen a brief or skeleton, they’re not skeletal or brief. And a little note of advice, because you may never been to court before. So things like put on a suit, turn up at court on time, address this judge by this particular form of words.
And that’s the basic idea behind Garfield. It’s supposed to sort of hold your hand going all the way through the court process. The other thing that’s really relevant is that I haven’t, with Dan and our team, who are completely awesome, and I will just talk a bit about them in a moment if you’d like, I haven’t just built this for non-law firms. I’ve also very much built it for English law firms to use as well.
as a sort of white label product because I’m really keen to create a platform that gives people choice because lots of people when presented with an AI product they might say I’ll give it a go but lots of people also say I want to use a human overseeing solution so we built it for both ⁓ and we’re very interested to see actually so far the take-up has been from both sort of wings of the both types of businesses.
Greg Lambert (18:49)
Yeah,
and I’ve seen where, know, all of this, ⁓ depending on, I guess, the level of service needed can be as little as two pounds or about $2.50 here in the US. At what point in this is there any kind of human interaction to kind of verify
what’s going on behind the scenes or someone has a question, ⁓ is there some way for a human to kind of verify or to talk to them?
Philip Young (19:28)
So one thing about being a lawyer, particularly a lawyer for a long time, is it makes you cautious, I think. Very rare to meet an uncautious lawyer. And I’m no exception. And I think with AI and any sort of new product, you’ve got to be both responsible and accountable. Those are sort of the two key themes, I think. And so as part of the process of becoming regulated, I was very keen to… ⁓
with the SRA that we would roll out in a measured fashion. We wouldn’t just open the doors and let the floodgates roll in and also that I would at the beginning stages sort of oversee all of the cases going through the system. Now in the longer term of course that’s not realistic and the plan is
to move, as the regulators also said, to move eventually to a sort of sampling process. Because a lot of documents in small claims are either template driven or the possible range of the possible universe of types of document is at least confined. It’s not infinite. And that means that you can eventually, with a large enough data set, apply analytics and mathematically examine the accuracy of your document against what you’ve received. And so over time, I will move very much to a… We have this mantra in England and I’ll…
in all of our regulators where we talk about risk-based approach to things. And so I will move very much to a risk-based form of supervision where I will focus my attention primarily on the parts of the case that are the most risky. In other words, a defended case where you actually have a defendant who is fighting. And so that’s a longer term plan. But for the time being, at least I’m checking everything. Obviously, I have been building this product for two years. So there’s been an awful lot of testing along the way as well. And I wouldn’t.
launched it unless I felt that it was ⁓ of an acceptable level of quality at this point in time.
Marlene Gebauer (21:25)
So, so, so now you have given me like three, three questions just from what you said, just from what you said. So the first one is, you you mentioned that, that you’re getting interest from, ⁓ you know, both like pro se people and also from, from law firms. So, ⁓ how, how is the marketing different for, for each of those? You know, how do you, I mean, I probably have a pretty good idea about the firms, but, how do you sort of get in front of.
sort of the regular folks that would be using this.
Philip Young (21:58)
Well, we weren’t actually expecting to be in the public domain at this stage. We weren’t actually expecting the FT, the Financial Times in London to pick up on this and to publish about this last Monday. And since they did, as I have mentioned online, my phone has melted with a sheer number of messages and emails. And I’ve been trying to sort of manage my way through the last week and a half whilst also continue to do my day job, so to speak. So marketing hasn’t really been at the forefront of my mind.
and I haven’t needed to really market. I haven’t had much time to do that. At some point I imagine we will do some proper marketing but I think it’s sort of early days still.
Marlene Gebauer (22:28)
Don’t need it. The Financial Times did it for you.
Philip Young (22:40)
What’s been amazing is the reaction actually of the profession in England and Wales and also more globally that people have been incredibly positive about this sort of development for legal practice. I think if I had any concerns before about the fact that we lawyers, we’re not renowned, are we, as a profession for being incredibly modern and forward thinking. And I can think of examples in my career when I was in front of high court judges up to 20 years ago. they would ask, they were trying to analogize a computer
with a quill because they’ve never used a computer and had no idea how it worked and now we’re moving into a world where even judges can use things like Excel and can manipulate PDFs and I think it’s quite fortunate actually that it means that the profession is a lot more sort of forward-thinking and willing to embrace things.
Marlene Gebauer (23:27)
Now my next question is, so Garfield AI, how far does it take the claimant into the process? Is it just sort of getting, excuse me, getting to court? ⁓ Or does like after judgment occurs, does it also give advice?
Philip Young (23:44)
So at the moment it goes up to the doorstep of the one hour small claims trial.
Interestingly, a lot of people have been talking about the fact that with the current rate of technological development, it may be possible for AI to do advocacy. I can see why that may become possible. have to say personally, philosophically, I’m not enthusiastic about AI doing advocacy and anything other than tiny claims because I think people should be represented by people in court because it’s more than just a process. It’s also a very human experience. Sometimes just because you need a lawyer there, because sometimes clients need a
shoulder
to cry on frankly. And that’s why we’re all amateur psychologists in a way aren’t we? That’s as far as it’s got at the moment. I think we will build in the judgment enforcement module. As it happens in England we’ve got a process underway at the moment to try and modernize the enforcement of money judgments so it will sort of slot nicely into that I think because there’s a desire to digitize that a lot more.
Marlene Gebauer (24:32)
That’s what I was thinking.
All and then the final question that I had from ⁓ what you were saying before is you were talking about analytics. so you’re looking at it in terms of risk and sort of where there’s argument, and I think that would be very valuable. Are you also looking at it to share with the courts in terms of types of ⁓
types of claims that are coming through, what the resolutions are, things like that.
Philip Young (25:17)
I would like to be able to do that, yes. I think one of the things I’ve been very conscious of, particularly in the last week and a half, is that because we are the first, we owe a sort good citizen obligation to the powers that be, so the judiciary…
political establishment, those sorts of people to try and help. And so as Garfield develops and as we end up with data, then subject to my professional rules around client confidentiality, if I can help them with data to help them deal with bottlenecks and things in the system, then I would be delighted to do that. The Ministry of Justice in our country does have reasonably advanced computer systems, not as advanced as they would like.
And there’s a lot of work being done to try and make them a little bit more 21st century capable. But they therefore do collect quite a lot of data already and publish it. But I don’t think they collect as much data as they would like to do. And products like Garfield will help.
Greg Lambert (26:15)
Yeah, it’s always good to make data-driven decisions.
Marlene Gebauer (26:18)
Exactly, exactly.
Philip Young (26:19)
Yeah exactly, yes,
rather than emotional driven decisions.
Marlene Gebauer (26:24)
But let’s talk about the SRA because ⁓ maybe some of our listeners don’t really know much about the SRA. I mean, I know for sure that I do not. But I know that getting SRA approval is no small feat, especially when you’re the first to ask them to approve an AI-based law firm. So tell us about the process. What were the guardrails? What were the controls that you needed to put into place to meet the regulatory standards?
Philip Young (26:52)
So yeah, let me just give you bit of context first. So as many of you guys will know, England has a split legal profession. So we have barristers who are specialist advocates, and they are regulated by the Bar Standards Board. And they have all these delightful ancient traditions that Americans in particular tend to like. So for example, to become a barrister, you have to eat a set number of meals in the Inns of Court in London.
So you’ve got to demonstrate the ability to eat, which is obviously a helpful ability if you want to keep living, and other things as well. There also has to be some legal ability. Whereas my side of profession, we’re known as solicitors and we have a different regulator, the Sisters Regulation Authority. If you go back in time, we were once known as attorneys as well, but the title was changed and I was told years ago it was because the title of attorney in England had fallen into disrepute and we were basically rebranded. Now I don’t know if that’s true or not, but that was what I was told.
an older lawyer. So the SRA, the Sisters Regulation Authority, that regulates law firms and it regulates solicitors and sisters work in law firms. And that was what I was and am for 25 years.
Early on in the process of building Garfield, because I used to have one of my side areas of practice was professional discipline and professional negligence, I’m pretty conscious of regulatory law around my profession and I thought to myself, whatever it is that Garfield might be said to be doing, I think it is conducting litigation. And if it’s conducting litigation, then that is something called a reserved legal activity in our Legal Services Act, which means you have to be regulated. So then I thought to
myself,
although I must go and see the SRA. So when we had an advanced prototype, I did two things. First, I went and saw the senior judiciary to make sure that the judiciary in our country were in favor of this sort of idea, because they cannot in any way endorse any particular product, but they can endorse the concept of products and say whether or not they like it or not. Because of course, if the judiciary hated it, then I’m wasting my time. I should go and do something else. Maybe my garden needs more work.
⁓ I went and saw the judiciary and they were in favor of this sort of thing as a concept and then I saw the SRA. And so the first question the SRA had to ask themselves was did they have any jurisdictional basis to even regulate this? In other words, is this new thing something that is a reserve legal activity? And they weren’t much assisted by our Parliament and our case law on that because Parliament’s never troubled itself to define what is conducting litigation and what is not. And case law has therefore struggled with particularly odd sort of fact patterns in trying to say whether one particular
odd thing was conducting litigation or not. But fortunately there was enough case law on enough of what Garfield did that I was clear that we were conducting litigation and the SRA reached the same conclusion. Having established themselves that they had the jurisdictional basis to regulate us then the question was well should they? And over a sort of thorough eight month process they asked ⁓ many many questions as was right and proper. They wanted to know about the people behind
it, where they fit in proper people to be regulated to run a business like this. They wanted to know about the product. They have a public duty to try to promote access to justice and so they wanted to understand how this would promote access to justice. And then they wanted to understand how this new type of law firm would work and whether we would fit within the existing regulatory rules and also our other ethical rules as a lawyer. And in England we have, I’m not sure if this applies to American lawyers,
came
across this when talking to American colleagues when I was doing parallel litigation between our two countries. But in England, solicitor is not just somebody who owes duties to their client. We also owe separate parallel duties to the court because we’re an officer of the English court. So we also had to look at that and were those duties in any way affected by this sort of product. And that ultimately was a big discussion about responsibility and accountability. And because I think I am a very responsible person and I’m certainly accountable, I think in the end
reached a conclusion that this and us were an appropriate thing to regulate. I should say though that the SRA had…
Greg Lambert (31:07)
Yeah. But just to answer your question
real quick, yes, lawyers here are also officers of the court so that they do owe a duty to the court and to the Constitution. And to the Constitu- yeah.
Philip Young (31:19)
And to the Constitution, oh that’s excellent. We don’t
go quite that far but that’s because we do have one but no one’s bothered to write it down in one place. You have to sort search around and it causes first year law students at university a degree of insanity to work out where it is. Yeah I was going to say one thing about the SRA is it’s very much an evolving process so we’ll regulate and authorise but we’re going to have regular catch up sessions with them where we share experiences and tell them how we’re doing.
are there any sort of bumps in the road that we’ve hit because we’re not just the first we’re also the guinea pig I think and that’s also a useful role for us to be.
Greg Lambert (32:00)
You had mentioned earlier, Philip, about ⁓ doing some white label arrangements with law firms. ⁓ Can you explain first a little bit about ⁓ what white label arrangements means and what kind of actions that can allow? ⁓ And then also, I’m curious. know… ⁓
with the AI tools, the legal industry is embracing these a little faster than we have pretty much any other technology, but there’s still a lot of naysayers out there. So I’m curious to see what kind of reaction you’re getting from your peers in the legal community there about announcing that this is an AI driven law firm or an AI law firm. ⁓ So white label first, ⁓ AI second.
Philip Young (32:51)
⁓ Yes.
Okay,
lovely. Yes. Okay. White labeling. So we’ve built the platform in such a way that, as I say, it follows the process of the way a small debt claim can go. And we’ve assumed that lawyers would like to use it as well. And indeed that’s proven to be the case because law firms have signed up. And what we’ve done is we’ve also assumed that law firms would prefer that when it generates…
documentary output that they have it on their letterhead, that they send it themselves and that they don’t have Garfield as a sort of external facing thing. So we’ve therefore set it up in such a way that people can put their letterhead on the system and generate the letters on there.
and I think the only thing we actually do that is external facing is that we because we are hooked up to the English court system we will send the claim form in the particulars to the court on behalf of another law firm and so there is about that’s about pretty much it that we do that’s external so in other words it sort of sits behind the scenes and the lawyer drops documents into it and takes documents out if a law firm said to us we want Garfield to send the stuff to then well we just we’d implement that I mean if that’s what they wanted but I’d be surprised ⁓ I always thought in practice
that all of us when we were in practice that there’s a certain amount of sort of professional pride isn’t there and ⁓ you know if it’s on your letterhead it’s yours and if it’s on someone else’s definitely not. So I think that’s white labeling. I’m sorry I’ve lost track of the other question.
Greg Lambert (34:18)
And then
how’s it been just the general reaction pro and negative on what you’ve announced?
Philip Young (34:26)
I did worry when it came out in the FT that I’d be taken behind the proverbial woodshed and shot.
That hasn’t happened yet otherwise unless I’m an AI avatar talking now. ⁓ No, so far the… It is possible yes. Probably do a better job than me frankly. So yeah, so far the reaction has been very positive and I’ve been really grateful for all of the kindness that people have sort of said in all their comments. I think people can see that this is coming and it’s going to happen.
Greg Lambert (34:38)
Yeah, it’s possible.
Philip Young (35:00)
that AI is going to lead to different ways of working. And the interest is really, well, what’s it going to do? And is it going to not just help the big law firms, all the big resources that they can pour into this, but will it help the ordinary people? And also the smaller high street law firms that do not have the ability to compete with the magic circle and silver circle firms of this world. And so I think that has meant that it’s received a very, very warm reaction so far. That said, it’s early days. Who knows? But so far, so good.
Marlene Gebauer (35:33)
So we talk about AI improving productivity a lot. But Garfield is directly addressing a multi-billion pound access to justice gap in the UK. So how are you thinking about that impact? Or what impact do you want it to make? And what stories are mattering to you most as you grow this business?
Philip Young (36:01)
So I would love to see people like Andy use it and collect in money and their life just be that noticeably bit better. I mentioned the summer holiday, they’ve got the summer holiday otherwise they wouldn’t have had.
and then I’m going to feel that if what Sister Mary said to me when I was a young boy going to ⁓ lessons after mass is remotely accurate, at least I I can plead all this in mitigation in front of the pearly gates. We’ll see. So yes, I hope that it will help.
sort of business ⁓ materially and I think as well it’s also a demonstration of the art of the possible so we’ve been able to do this other similar products will inevitably be launched and many of them I think will be launched at the consumer space or the small business space and so debt claims of course is part of the overall sort of iceberg but other parts include all sorts of other areas of law and I can think of ⁓ plenty of cases I’ve heard about over the years for relatively small sums of money but
really important for the claimants in question and they’ve not been able to pursue them. So if those sort of products come out then people are going to be, their lives are going to be enhanced by the fact that they can go off and have a judge tell them whether they’re right or wrong. Vindicated to use a word.
Marlene Gebauer (37:24)
I think it could possibly increase business for the attorneys that do this type of work because it would, as you said, sometimes people want to have a person and maybe they want a person in court to actually do the arguments, but they can get through a lot of the paperwork much faster and sort of move those cases much quicker.
Philip Young (37:48)
I think you’re absolutely right, Marlene. So I was jokingly saying to a friend of mine about a year ago, so there’s, I think Charles Dickens in Bleak House, that whole story is of course a slating of the then chance record system in England and Wales. And he made the comment that the business of English law is to generate business for English lawyers, which is obviously a very cynical comment.
Marlene Gebauer (38:12)
It’s just like America.
Philip Young (38:13)
There might be
some truth in the comment, who knows, wouldn’t want to comment on that. But I think that in answer to the concern of will AI take all our jobs, I don’t think it will. And I think it’s going to have two impacts. So I think back 25 years ago to when I was a junior lawyer and I had hair and it wasn’t gray anymore. And as a junior lawyer, you had to do an awful lot of document review and disclosure exercises and due diligence.
It was all very tedious, worthy but tedious. And what you really wanted to do was a higher level quality stuff that lawyers go into law to practice. thinking about strategy and tactics, dealing with clients, formulating the case, coming up with the arguments, finding the right law, those sorts of things. And I think that what will happen is that AI will take away some of the more mundane, tedious administrative tasks that are involved at the moment in the practice of law, free up lawyers to do the more interesting higher level
or higher value skills and simultaneously because it will reduce the cost of it certainly litigation and I can’t really speak to non-contentious work but I can speak to litigation. think what will happen is a lot more cases that presently can’t be pursued because they’re uneconomic will become rapidly very economic and that means that there will be more disputes and I think the other thing AI will do is it will enhance the overall quality of lawyering. So I can think of many lawyers in my years of practice who are excellent but I can also think of some who are
excellent, if I could put it like that. And I think a lot of bad lawyering will become ameliorated by the fact they’ll have an AI assistant on hand that will say things like you can’t put that in a letter, or that’s a really bad argument, or that’s a really good way to annoy the judge. And I think that will mean that the quality of lawyering will become higher.
more cases that should go further but fail because of negligence or incompetence, they won’t have that problem anymore. So I think the two factors balance each other out. I think there’ll be more work in terms of greater volume of work and the lesser attractive work will drop. And so I don’t think we’re gonna be, we’re not gonna be replaced just yet. And I’ve also said to people when they’ve asked me this question that certainly in a contentious, in the contentious area of…
I think the only circumstance in which lawyers would become obsolete and redundant is if human nature suddenly changes and everyone becomes terribly reasonable, honest, fair-minded, self-sacrificing, decent people. And that’s not really where the human race is at the moment. So there will always be disputes.
Greg Lambert (40:47)
Yeah,
it’s interesting. I want to pull on the thread of something you said earlier, and that is that we have the potential to mitigate what you refer to as potentially bad lawyering. Whereas before, when we talked about efficiency and scaling, you know, if you…
if you take a bad process and you make it more efficient and scale it, it’s just a bad process at scale now. Where I think we’re finally, since this type of tool is something that when used correctly can actually improve ⁓ the overall output ⁓ and process, I think this is kind of a unicorn effect of where it could actually, like I said, improve.
potentially and I think bad lawyering may be a little too harsh but I mean just kind of expand the ability to do more good for a client in the same amount of time so ⁓ thanks for pointing that out.
Philip Young (41:54)
Exactly yeah I mean
I was jokingly saying to another lawyer that I’m surprised that no one has yet built a very simple AI tool that just goes through your draft documents and removes all the adjectives and then understates everything because that’s the first thing senior lawyers do to junior lawyers. Out comes a red pen, no adjectives you know. There you go, if you make a lot of money remember me.
Greg Lambert (42:13)
All right, Marlene, there’s our business plan. yeah. Yeah, Philip, we’ll go to
Marlene Gebauer (42:13)
That’s a great idea. That’s a really great idea.
Greg Lambert (42:21)
the VCs right after this call and pitch this idea. So I’m going to give you kind of a hard question because I know you’ve had a good three months of experience with your trip here in the US. But access to justice, even in small claims courts, is something I think that
Philip Young (42:23)
Ha ha ha ha ha.
Marlene Gebauer (42:33)
You ⁓
Philip Young (42:34)
Ha
Greg Lambert (42:43)
The United States is known for really falling down on that and I think there’s statistics that over you know potentially over 80 % of legal work goes unmet Just be you know because of the the process the cost and other things so we have a huge access to justice problem so if a US lawyer were to come to you and say do you think this is something that we could you know create here in the US or
another country, maybe another common law country, do you think this is something that’s replicatable in other countries?
Philip Young (43:23)
Yeah, think philosophically, totally it is. And actually one of the consequences of the huge amount of media that’s fallen upon us in the last week and a half is that I’ve been contacted by senior judges in other countries who have found me through friends ⁓ and other contacts and said, you know, can we talk about this?
because I think everyone’s interested in trying to improve access to justice. So I think there will definitely be this sort of product rolled out in other countries. I think, however, the scale of the challenge should not be underestimated. mean, it has taken us two years and an awful lot of thought and cold towels on our heads to build what we built. And also we built in a very benign environment because the political establishment in England is very pro-AI. We’ve got very much a government with a growth mission and a pro-AI mission at the moment.
very important. Not sure all countries have got that. Secondly, the judiciary in England are incredibly forward-thinking and whereas some countries may not have that, we have a judiciary that has been willing to say that they would like to see products created and deployed that maybe some countries wouldn’t allow.
Thirdly, we’ve got a regulatory framework that has worked to enable it, and not every country may have that. And by that, I mean the legal regulatory framework.
as well as sort of more general law. ⁓ And fourthly, even things like insurance, because obviously if you’ve got a law firm, you’ve got to get insurance. And insurers are notorious for wanting to take on novel risks as a business class. And so there’s that as a challenge. And then I think as well, you’ve got to have a court system that has the technology that enables you to connect to it. And not all countries have got that.
We are connected in England to our bulk issuing center in Northampton. ⁓ Their technology is acceptable, but even they would say that they would like it to be about 10 years more advanced. So I think you’ve got those sorts of challenges, and then you’ve got to build the product. So that’s all of that, and then you’ve got to build the product. And the product is very much a non-trivial thing to build, I can tell you. It’s taken a lot of very careful thought and an awful lot of testing to get it to the point where it is.
Marlene Gebauer (45:43)
Philip, I want you to pull out your crystal ball. So in the next two to five years, how do you see AI changing the way laws practiced? So not just for consumers and small businesses, but for the legal system as a whole.
Philip Young (45:56)
Okay, the first thing I’d say is that I’m notoriously bad at predicting the future. When I was in practice for many years, people would sometimes ask me, you what’s the next year of CYK’s business going to be like? And I would always be wrong. Sometimes I’d say it’s going to be the same, sometimes it’s more, sometimes it’s less, sometimes I’d it’d be more and it was less. So you can’t, predicting the future is incredibly difficult. I think if I was going to go out on a limb, what I would say is that…
the uptake of AI and the progression and scaling and deployment of AI products is just going to accelerate.
And the art of what is possible is becoming more and more open all the time. people are thinking of increasingly novel products. And I think in the legal space, it’s very much a question of ⁓ when not if. AI is there that sort of is an AI assistant that supports you as a lawyer or even one step further and is a sort of thing that we’ve built where it is also able to conduct a particular matter by itself. What I would say is that no one should panic because I’m very conscious of the current state
of technology and I would not myself feel comfortable where we are at the moment with trying to handle much bigger claims because I don’t think the technology is nearly capable enough for that because the the complexity of cases is not a linear increase I think it’s a net more of an exponential curve as claims get bigger and more complicated there’s more parties and so I think that the technology would have to become exponentially more capable to go up the value chain
So that’s where I am at the moment. think if you’d asked me three years ago, what would you be doing in three years time? And then you’d said to me, you might be the CEO of a law firm that’s just launched an AI product. I would have laughed and laughed and laughed and said, whatever you’ve been drinking, it’s good stuff because there’s no chance. So I mean, that’s why I can’t predict the future. Who knows? But I think it’s going to be a very interesting time. And I think if I was a junior lawyer, I would be not just working on my legal skills, I’d also be working very much on my
Marlene Gebauer (47:41)
you
Philip Young (47:56)
technological skills because I think they will hold you in good stead going forward.
Greg Lambert (48:01)
Yeah, we just our summer associates just started this week. So I will take this clip of you saying that and send that to them. So
Philip Young (48:07)
In that case
also do everything Greg tells you because you’ll learn nothing.
Marlene Gebauer (48:11)
Hehehehehe
Greg Lambert (48:11)
Exactly. ⁓ Yeah.
A ⁓ 20 pound note is coming your way. Thank you. ⁓ Well, Philip Young, co-founder and CEO of Garfield AI. ⁓ Thank you very much for taking the time to talk with us on the Geek in Review. This has been fun.
Philip Young (48:32)
Thank very much. Thank you for your invitation. It’s been very nice to speak to you and ⁓ also to sort of have a Texas to England connection here.
Marlene Gebauer (48:32)
Thank you.
And thanks to all of you, our listeners, for taking the time to listen to the show. If you enjoy it, please share it with a colleague or give us a shout out on LinkedIn or Blue Sky.
Greg Lambert (48:51)
And Philip, we’ll put links on the show notes, but if listeners want to learn more about you or about Garfield AI, where would you direct them?
Philip Young (49:03)
Well, the first thing I’d say is I’m incredibly uninteresting as a person, so you definitely shouldn’t try and look me up. But why not look up Garfield AI at www.garfield.law? There’s some videos there. There’s a speech that was given by Law Justice Burrce, who is the deputy head of our civil litigation, talking about how this sort of technology can really help the court system. So things to look at.
Marlene Gebauer (49:28)
And as always, the music you hear is from Jerry David DeSica. Thank you, Jerry.
Greg Lambert (49:32)
Thanks Jerry. Alright, talk to you later.
Marlene Gebauer (49:34)
Okay, bye.
Philip Young (49:35)
Bye.