Want to listen to the full episode and all our other episodes?
Hearsay allows you to fulfill your legal CPD requirements every year.
Our yearly subscription is only $299/year.
With a yearly subscription, you can access all of our episodes AND every episode we release over the next year.
Episode 116
Buy Episode
Practice Management and Business Skills Professional Skills
Democratic Revolution: Levelling the Access to Justice Field Using AI
Law as stated: 16 February 2024
What is this?
This episode was published and is accurate as at this date.
Entrepreneur and CEO of Lawpath, Dominic Woolrych, joins David to talk about the democratising power of AI on access to justice. Touching on access to justice as a small business issue, using AI to scale legal services, and finding the right business model.
Practice Management and Business Skills
Professional Skills| What area(s) of law does this episode consider? | The democratising power of AI in law |
| Why is this topic relevant? | We’ve spoken on the podcast quite a bit this season about the potential of AI to transform the law and legal services. We’ve looked at the regulation of AI with Raymond Sun, some practical use cases for lawyers with Fiona McLay, as well as the impact of the technology on the teaching of law with Jason Harkess. Today we’re looking at AI in legal services from a broader perspective: the power of AI to democratise law. That is both the provision of legal services, and access to information. |
| What are the main points? |
|
| What are the practical takeaways? |
|
DT = David Turner; DW = Dominic Woolrych; RD = Ross Davis
| 0:00:00 | DT | Hello and welcome to Hearsay The Legal Podcast, a CPD podcast that allows Australian lawyers to earn their CPD points on the go and at a time that suits them. I’m your host, David Turner. Hearsay The Legal Podcast is proudly supported by Lext Australia. Lext’s mission is to improve user experiences in the law and legal services and Hearsay The Legal Podcast is how we’re improving the experience of CPD. On this season of Hearsay, we’ve talked quite a bit about the potential of AI to transform the law, the legal profession, and the provision of legal services. In this season, we’ve looked at the regulation of AI with Raymond Sun, some practical use cases for lawyers with Fiona Maclay, and the impact of the technology on teaching law with Jason Harkess. Today, we’re looking at AI and legal services from a broader perspective, and maybe a more hopeful one, the power of AI to democratise the law and access to legal information. That’s both the provision of legal services and that more general access to legal information that doesn’t rise to the level of legal services. Now, our guest today is well-suited to that discussion, Dominic Woolrych, the CEO of Lawpath. Lawpath is a legal services platform that allows users to easily create legal documents, manage compliance, and connect with lawyers on demand, and it’s recently launched a suite of generative AI-powered tools as well. I don’t think there’s anyone better to shed light on how AI is reshaping the accessibility, efficiency, and affordability of legal services in Australia today. Dom, thanks so much for joining me on Hearsay!
|
| 0:01:47 | DW | David, thank you so much for having me on the podcast. It’s a great privilege to follow some of the names I just heard, Raymond, Fiona, Jason, all distinguished guests, so I’m very happy to be in that cohort.
|
| 0:01:59 | DT | Yeah, we’re kind of seeing over the past year all of these names emerge as kind of the experts on this nascent field, right? So it’s been great to speak to all of them. I’m looking forward to our discussion today as well. We’ve had a few of these conversations already, albeit not in the recording room, I suppose. Both Lawpath and our company, Lext, have AI-powered legal productivity tools, although I suppose they’re slightly different audiences. We’ve talked a lot about the potential for these tools, both in the legal profession and for the public at large, so really excited to jump into it. But before we do, I wanted to get you to share with our audience your journey in the legal profession, sort of how you got to Lawpath, how you got to the role that you’re in today. You were originally a lawyer?
|
| 0:02:45 | DW | Well, these days I consider myself a recovering lawyer! But yes, I started my career as a lawyer. Very traditional, went through a double degree at University of Technology here in Sydney, and then a clerkship at MinterEllison, and then joined the corporate team at MinterEllison as a commercial lawyer, and really loved that. Loved my time there, spent three or four years there. And I think when I was there, I started to see the impact that technology could have on law, or maybe better said, the impact that technology wasn’t having on law. And I felt that was a really big opportunity. My dad is a small business owner, and he would constantly come to me and say, “Hey, Dom, I’ve got this legal problem. Can Minter Ellison help me?“. And I’d say, “Dad, we can probably help you, but you probably can’t afford the fees that are gonna come off the back of that help“. So I started to notice that there was a big gap in the legal industry around access to lawyers. And I think really interestingly, we normally focus on the gap being more at the level of personal law and legal aid and the access there. And what I felt over those couple of years was that small businesses still had this issue where they couldn’t access legal services, but because they were a business, it was assumed that they could.
|
| DT | Yes.
| |
| DW | And so there was this big gap in the middle there whereby there were a lot of small businesses just foregoing the protection that they could get from engaging with a lawyer or engaging in the industry, and you’re just going it alone. And I felt that it was a big opportunity there. So in 2014, made the leap from Minters to Lawpath, which is the company I’m CEO of now. And really, we wanted to, I’ll be totally frank, when we started the business, we didn’t know what we wanted to do. We knew at a very high level, we wanted to introduce a model into the Australian legal system whereby we use software, and leverage software, to provide legal services, first of all, cheaper, and then provide access. So there were a couple of key tenants of the business model. We knew it needed to be online. We wanted it to be as self-serve as possible. And we knew that it needed to be cheaper than the current offerings. And it probably took us three or four years of trying different products and different revenue streams to find the kind of the secret sauce of what Lawpath is today. When we started the business, the very first model was a marketplace. So we actually used to share an office with Airtasker. So I’m not sure if you’ve ever got your… you’ve moved house and you needed someone to build your bed or build your TV cabinet, you’ve probably used Airtasker. And we thought, we can be Airtasker for lawyers. We can build up a system, an online marketplace where you can engage a lawyer online. And that worked really well for a little while and lawyers loved it because it gave them access to new customers. We were national, so a lawyer in Melbourne could work with a client in Brisbane. However, we kept getting these feedback from clients saying, “look, it’s great that you found me a lawyer, but I actually could do this myself or I want to try and do this myself, but I don’t have the knowledge and I don’t have the tools”. And what we saw was there was this big opportunity to empower small businesses to complete certain legal tasks themselves if they had the tools available to them. So we totally pivoted our model and we started building a platform, we started building software that allowed small businesses to complete really simple legal tasks themselves. Without the need of a lawyer. And really over the last six years, we’ve been building on top of that. And really where we’ve landed is there are certain tasks that a business needs to complete that they can do themselves and given the right tools and the right guard rails, and I’m sure we’ll talk about guard rails later on in the conversation, but given the right guard rails, we want to give them those tools. Let me give you a really simple example – an NDA.
| |
| DT | Yep.
| |
| DW | One page NDA, super standard, used software. We don’t need a lawyer to custom draft that. All the lawyers listening to this know that they are relatively straightforward documents.
| |
| DT | Yeah, I mean there’s even projects internationally like OneNDA that are aiming to make a uniform template for NDAs to sort of finally put to rest the custom drafting of pretty standard documents.
| |
| 0:07:15 | DW | Exactly. So that’s one end of the spectrum. And then the other end of the spectrum are there are most definitely legal tasks, legal projects that go on where you must have a lawyer involved. It’s complex. So many things can go wrong. The risk is too high. So the challenge was where’s the line? Where do you allow someone to do it themselves? And where do you say, “you know what, we want to step in here for kind of consumer protection reasons and say you need help”. And so now we’ve got to a point on our platform where you can; there’s tools to do it yourself or there’s ways to really easily engage an online lawyer and depending on what you’re doing you kind of flip between those ones.
|
| 0:07:53 | DT | And you kind of have a vertically integrated solution for accessing the advice of a lawyer if the customer needs it, right? You have that in-house.
|
| 0:08:01 | DW | Yeah, exactly right. So we operate a law firm as well and then we also operate a network of law firms around the country. So you can choose depending on the level of service you require what solution works best for you. Probably going back to, I think I probably spent more time in the legal industry now as an operator or a technologist than a lawyer, but I still think that we are in an industry which is technical and you need to stay relatively close and on the tools. And so I still do pick up the occasional piece of legal work and do that. But I’m very much passionate in building tools that allow us to transform legal from a one-to-one medium to a one-to-many. You know, you have one lawyer, you have one client traditionally, they provide the advice, they walk out the door, the next client comes in and it’s always one-to-one. And what that means is you can’t scale above and beyond your capacity as a lawyer – 60 hour week if you’re working hard. So, but I’m really passionate in building software that allows us to provide the software to multiple people. And we’ve got now over 450,000 small businesses on the platform and there’s, you know, you could only manage that kind of volume with software.
|
| 0:09:15 | DT | Yeah, absolutely. A few things I want to unpack there. First, I really love your description of access to justice as a small business problem. You know, I think when we think about access to justice, we often think about it through the lens of, I suppose some of the more traditional solutions to access to justice. Legal Aid, the community legal center, sector who do terrific work for individuals who have criminal or family or other personal law issues. But we don’t often think about the kind of small business access to justice issues. But we know that I think about 85% of demand for legal services in Australia is unmet by the legal profession. A lot of that is civil law. A lot of that is business or commercial law. And I suppose thinking about that as an access to justice problem, at least for me, I had to kind of reframe what I thought of as a legal services problem. Like the sort of thing you would go to a lawyer about. The metaphor I sometimes use is that, you know, if you had to pay $600 an hour to see the doctor, the sort of things you would go to the GP about would be very different to the things we go to the GP about now, right? You’d probably wait until an arm was falling off before you went to see the doctor. But because medical advice is well subsidized by the government and mostly affordably available, we’re able to see the doctor about all sorts of non-life threatening conditions. And I think we have a relationship with legal advice at the moment where we sort of need the equivalent of the life threatening condition to actually get any advice about it. But that doesn’t really need to be what it is. And I think what we’re going to talk about today is, you know, you described that delineation of, well, there’s the low risk stuff that someone can do on their own. And there’s the high risk stuff that really needs the advice of a specialist. What artificial intelligence is doing is pulling the sort of closing bracket of that first category further along the curve to, you know, there’s a greater category of things that can be done on your own than ever before.
|
| 0:11:12 | DW | I’ll just jump in there. There’s two really important points I think you just made. So the first one around, as much as we may, as solicitors and lawyers, think that there’s not a barrier to our services, there is a barrier to our services. And that might be a location barrier, a time barrier, a cost barrier, but there are these barriers. And so you’re exactly right. You know, you hear people when you speak to small businesses, you say, “oh, you should speak to a lawyer to get help with that”. Like, “oh no, I only go to a lawyer when things go wrong, or I only go to a lawyer when I really need to”. And the parallel I really like to draw is to the accounting industry. So 20 years ago, you’d only go and see your accountant once a year to do your books, right? Now, as a small business owner myself, I log into Xero every day. And so what Xero has done is it has built a platform and removed all those barriers and all that friction so that I can then become more proactive with my accounting. And look, I still go and see my accountant face to face, but the majority of the work is done online in my own time using a self-serve tool. And I think that law has a really big opportunity there to do the same. And again, swapping this idea that legal is reactionary to legal can be proactive. And yes, the big corporates, they can have lawyers on retainer and those lawyers can be proactive, but a small business can’t afford that. So how do we do that? I think technology and scale fixes that. And then just to your point before that, which I thought was really interesting about access to justice, I have a little bit of a different take on a lot of the access to justice programmes. I mean, yes, the government puts in place some fantastic programmes and they can help. They’re not perfect, but they can help. I do think though that for software and for efficiencies to creep in to the access to justice space, often you need a business model that there’s a commercial driver for it. And then whatever is created there can trickle down into access. So a really good example of that is our automated will products. So we built our automated will products to be purchased commercial, and we drive revenue from them. Yet what we often do is provide them for free to charities or free to associations. And they can use that software to fix an access to justice problem. So I do think that what we see is the disruption or the innovations trickle down into places where they may be needed, but there’s no funding for them.
|
| 0:13:44 | DT | Yeah, I suppose historically, funding has been so tight in the not-for-profit legal sector that it’s kind of a struggle to continue to operate in some ways. There’s not a lot of money for CapEx.
|
| DW | Yeah, exactly.
| |
| DT | But I love your metaphor of kind of accounting and Xero as the transformation that we could achieve because one, I don’t think anyone thinks of their Xero subscription or MYOB or whatever cloud accounting software you use as a replacement for accounting advice. It’s not a profession killer, right? It’s a product that can only exist by the combination of software and accounting expertise. It provides the sort of information that 20 years ago you could only have got from your accountant or specialist accounting training yourself, but no one thinks of it as a professional service. It’s just a way of accessing information and people still go to their accountant for those less rote tasks, I suppose. You were talking about this one-to-one versus this one-to-many approach. It’s more scalable, sure. It’s a more financially rewarding business model. I think it’s just more economically efficient from a kind of position agnostic perspective. The same lawyer producing the same NDA 60 times versus producing it once for 60 use cases just makes a lot more sense. And I think it’s more fulfilling. I don’t know that anyone got into law to fill out the same NDA template over and over, right?
| |
| 0:15:15 | DW | Yeah, exactly right. We got into law because we wanted to solve problems and think strategically. And I think there’s still very much a place for that. What we saw in the accounting space is software came in and it disrupted the bottom. So yes, there were some bookkeepers whose task was just to reconcile bank accounts and accounts and ledgers. And yes, Xero took their job, but they didn’t put them out of business. It just meant that those bookkeepers could move up the food chain and start doing the tax or the other tasks that are required. And I see that, I hope that’s the way the legal industry will move. And I think there’s a lot of signs already showing that’s the way it will go, which is the software, the AI, whatever is coming in will likely take the bottom, kind of commoditize tasks. And what that will mean is lawyers can spend more time on the more personal things. We got a lot of negative feedback in the early days when we created our business from lawyers saying you’re ruining the profession, you’re taking our jobs, you’re replacing us with software. And that just, it’s not happening. So what ended up happening was we took away a lot of the work that just didn’t want to be done. I mean, one of the practical things I did a lot of, I went and interviewed a lot of small practitioners and one of the big pieces of feedback from them that blew me away was we don’t take on a job under $2,000 because it costs just too much to open the file and get it sorted. So if someone has a problem that is only a couple of hours of work, we’ll likely just not accept the work. That’s the majority of the issues that small businesses have. They’re just quick things that need to be solved. And so I think that we’re actually going after or software or AI is probably gonna have its largest impact on what I call the under-advised, which is all of those that actually are, we’re never going to engage a solicitor in the first place.
|
| 0:17:13 | DT | Exactly, yeah, like we said, 85% of demand for legal services is already unmet. I think those software-enabled or AI-enabled services are growing the pie, growing the market of served customers. They’re not really taking away the customers who are spending $50,000 a year on legal advice, right? There might be a small number of tasks that lawyers are currently doing but don’t particularly enjoy doing. that can be automated but more likely, these sort of direct-to-end customer products are serving people who were never being served by the traditional profession in the first place.
|
| DW | That’s right. | |
| DT | Now, we’re talking about AI in particular today. We should mention we’ve both got horses in this race. Little bit of overlap in some of the functionality, I suppose, around reviewing legal documents and things like that. But I guess as we’ve said a few times, plenty of room for everyone to succeed here. So we’ll try not to be too fueled by our rivalry here. We’ll try to leave that at the door. What is Lawpath doing at the moment with AI? How are you using AI to build on your existing product?
| |
| 0:18:18 | DW | So Lawpath has built a product called Lawpath AI, very creative naming there. And we, you know, our goal from the beginning has been to find a legal process that can be commoditized and as much as possible, augmented through software. And the recent developments in AI, especially around access to these tools, has meant that it’s really unlocked a lot of the kind of use cases that we always wanted to build. It’s unlocked the ability to do them. So we’ve launched six features so far. So the way that we thought about our AI system was instead of building it into one of our products, we built it as a layer on top of our platform so that it could jump into any of the products. But we wanted to choose, the first one we chose was document-based. So we have a document contract management system, allows you to create draft e-sign and store documents. And so we’ve launched draft and review. So the ability to draft letters, clauses, and simple legal documents. You know, I’m sure everyone has used a simple document builder before where it’s essentially a conditional logic system whereby you say, “do you want a non-solicit clause? Yes, no. Do you want to non-compete? Yes, no”. What the draft feature does is it’s plugged into our database. So the database has anonymized data on a couple of hundred thousand businesses. And so we know a lot about your business when you’re using our platform. And so based on that data, we can actually suggest the clauses or the customization that we think you need. So let’s say for an example, you’re an accountant. We know that a lot of accountants will come and customize our documents and put professional CPD requirements within the contract that says you must keep up to date with, or you must be registered as a tax agent, or whatever it might be. So instead of the new client having to think about that and typically they go to a lawyer and the lawyer would say, “oh, you know what? You should put this in your contract”. The AI system can look at all the other accountants that have created contracts in the system and say this is a really common clause that they’re using and then it will suggest it to you and say, “look, we think you should plug this in”. Now at the moment, it’s through more of a chat interface. Where we’re looking to go is that it will essentially load up the optional clauses that you might wanna put in in like a traffic light system. Say, “yes, I’ll have this one. No, I won’t have this one”. And maybe go as far as saying nine out of 10 accountants in Sydney also had this because one of the learnings we had even before we implemented AI was people don’t want options. They want the confidence that the option they’re choosing is correct. And that’s really counter to what lawyers want. Lawyers want a clause bank of 15 clauses and so they can say, “you know what, this is the best one. I’m gonna use my judgment to choose that”. Small businesses don’t. What we found when we put 15 clauses in front of them, they thought that’s just more confusing. What they wanted was one clause.
|
| DT | That’s social proof.
| |
| DW | Exactly.
| |
| DT | Yep.
| |
| DW | And I think data can really help with that because if we know that a 1000 accountants created an employment agreement and 990 of them put this clause in, that’s, you know, we should be putting that clause in. So draft review, so document review will, it’ll essentially review a Word document or a PDF or a document that you’ve created on the Lawpath platform and it will read it and pull out key terms. It’ll also suggest additional clauses you might want. It will also look and see if there’s any missing clauses from market, from like a market standard. Still working on that one. It’s not like perfect yet but it’s definitely come a long way. And one of the use cases we’re finding there, which is really fun, is that – fun when it comes to AI, but we have a lot of e-signatures that go through our platform, DocuSign, HelloSign, LawpathSign, if anyone’s out there looking for a new e-signature solution. But we pulled some data and we found that the average time people spent reviewing a document before they e-signed was something like 15 seconds.
| |
| DT | Sounds about right.
| |
| DW | And there is no way that you are reading a full legal document in 15 seconds. And so what we kind of surmise from that is people are accepting these documents and they’re just signing them. And we know that you signed a document, it’s legally binding, you could be agreeing to some things that are really not okay in there. So what we’re actually using our review feature for is when you get your e-signature request, it actually summarises the document for you and says like before you sign this, here are the parties, here are the key terms, here are the risks.
| |
| 0:22:53 | DT | I saw a post on LinkedIn about this. I think this is a great idea because it’s a refreshing dose of reality, I think. As lawyers, we sometimes live in a bit of a fantasy land where like; “well, everyone should read what they sign” and the reality is busy executives aren’t going to read the 50-page document again before they sign it. Usually they’re expecting that whatever’s been sent to them is the correct up-to-date draft. And so it’s a bit of a concession to reality, I think, to say, “look, we know you’re not gonna read the whole thing, here’s the dot points you should really read before you sign”.
|
| 0:23:27 | DW | And look, it’s a relatively light implementation of AI, just summarising a document and giving the key points, but it’s a real-world, like, rubber-on-the-road use case that we thought was really useful for people. I think just to your point around reading terms and conditions, I heard this amazing story the other day of a SaaS business, a subscription business in the US, actually buried in their terms and conditions; “if you read this and email this number, we’ll give you a free account”… and no one emailed them for four months! So it just showed you that no one reads the terms and conditions.
|
| 0:24:00 | DT | Ever since I heard about these Easter eggs, you know, I can’t remember the company that did it, but there was one that said that they, you know, that the terms and conditions included selling your firstborn or your soul or something to the company. I look for these Easter eggs now, and unfortunately the search often doesn’t show up anything good, but it is funny to hear about what people bury in there, because it’s the reality, right? We kind of all accept it, even as lawyers, I don’t know that we always read them. My perspective has been for a long time, since being a relatively junior lawyer, was always; “well, if I can’t negotiate the terms, I have to think, is there anything in here that’s gonna stop me from clicking accept?” And if not, I just should click accept.
|
| 0:24:44 | DW | I think it’s always funny when you’re watching your iPhone load the new software and you have to agree to the terms and conditions and you’re sitting there thinking, what am I not gonna agree to these?
|
| 0:24:53 | DT | Yeah, I can either agree to these terms or throw the phone out.
|
| 0:24:59 | DW | Exactly, so draft review, we’ve got translate, so we can translate legal documents into a number of different languages, simplify and define terms. So again, document-based, it will simplify clauses for you, it will translate clauses for you. We found a lot of the time, we did a bit of analysis of what our lawyers were speaking to our clients about, kind of categorise each of the consultations, and a large portion was just, “can you explain this to me? Can you explain this clause to me?”. So the AI, as a first step, will try and simplify and explain the clause to you. Now, a lot of the time, we’ve actually put some guardrails in there that says, “hey, this is what, we’ve simplified it for you, but this is risky, you should also get a lawyer to explain this to you”. Some of the more exciting features that we’re launching at the moment, which are actually more under the hood, behind the scenes, is recommendations. So just like you’re in Netflix, and suddenly all the TV shows that you want to see start popping up at the top, we’re actually starting to apply that type of system to law parts. So what we found was that often clients just don’t know their next step. So let’s say I’m setting up a brand new business in the fintech space, what do I need? What licences do I need? What registrations do I need? What documents do I need? And typically you would go to a lawyer, maybe a financial services lawyer, and they would say, “here’s the roadmap of the next six months, we need you to get an ASL, we need you to do this, we need you to get this insurance”. So how can we take that and actually use AI to provide that? So through the platform now, it will suggest the documents you need, the registrations you need, the permits you need, again, based on the… so we call our database a cortex, essentially – and based on what people have chosen in the past. And then the last one that I’m quite excited about is, the part of our service is that it’s a subscription fee. So our model is not an hourly billing or a kind of a one-off billing, it’s a, you pay a subscription fee, you pay a monthly or an annual fee to be part of the platform, depending on which tier you’ve chosen, you get access to different things. Our most popular tier is you actually get access to on-demand lawyers. So if you have a question or if you’ve got a concern or you’re doing something on the platform and you’re like, “look, I’m just not 100% I need to speak to a lawyer”, you can actually live chat, you can call or you can, they’ll pop up on a video screen. And what we found is these lawyers having tons and tons of conversations every day and naturally now we’re using AI behind the scenes to actually prepare the lawyer for those conversations. So the AI will summarise all the actions that the client’s taken, summarise all the documents they’ve done, their company, their cap table. And so when the lawyer gets on the phone with them, they don’t have to ask any questions because they have a full summary of what the client has been doing. But better than that, typically we know the question that the client’s going to ask and the AI’s actually provided the answer. So it’s there, you’re actually seeing another use case, which is this kind of co-pilot or assistance for lawyers, which I know a lot of, that’s probably the key use case at the moment in the industry for AI, which is an assistant or a co-pilot for the profession. Rather in our case, what we’re doing is saying, we want to provide the efficiencies, let’s call them, to the end user, to the actual client.
|
| 0:28:24 | DT | Yeah, well, I want to talk about use cases for lawyers in a sec, but something that’s come up and I’d be really interested in your perspective on this, both because you’ve been kind of confronting this challenge long before AI came on the scene or came on the scene to this level of ubiquity, but also because it’s one we’ve talked about before. The old refrain when we were providing a digital legal solution that didn’t rise to the level of legal advice, which you needed a lawyer to sign off on, was this is general legal information, it’s not legal advice. Something that AI enables us to do, but it also presents new challenges, is personalize that general information like never before, right? The answers, the summaries, the descriptions of the legal documents are specific to that document. The answers to legal questions are specific to the way the users frame that question, including the hypothetical factual scenario they’ve presented. Where do you think we have to draw the line between legal advice and legal information and kind of how do you manage that risk at Lawpath?
|
| 0:29:35 | DW | Yeah, it’s a really good point, good question and good point, and I think something that the profession is going to need to tackle as these new tools come out. So a couple of ways that we look at it, I think the first way is that we state as much as possible when you’re using a tool that’s not a lawyer, that it’s not a lawyer and it’s not legal advice. That’s the first one. But we do know that disclaimers are not enough. If you hold yourself out as a lawyer or you provide legal advice, we know the legislation. So I think one of the other ways we look at it is we always provide two options. We always provide the software route and a human lawyer. And so if you’re using the software, you’ve opted not to use a human lawyer. And if you’re using human lawyer, it’s the inverse. So we have had to kind of pull back a little bit on some of our features to say we will only go this far up to the line. If you need more information, if you want more specific advice, if you need more details, that’s when we refer you over to a lawyer. And look, as long as you can get that process seamless, it’s actually a great experience because what it does is it gives the client some information upfront, gets them going down the path. They might realise actually this is the wrong path. They can rewind before they’ve spent a whole bunch of money. And then if you can seamlessly swap them over to a lawyer and give the lawyer all that context beforehand, at a net level, you save money. The very last thing that I think is really important is we’ve built in certain guardrails to our AI system, which stops it from doing certain things. So if the user is talking about certain areas or if the client is starting to get into an area of law which we’ve kind of deemed as high risk, it will often caveat anything by saying this is general information. I’m only going to give you general information. You need to speak to a lawyer. And I think the last one is always citing the source as much as possible. For example, in our contract review system, if the AI reviews a document, it will always point you back to the clause where that answer came from. And as much as possible, try and say, “I’ve pulled this for you, but here’s the clause, here’s the source, you need to look at that yourself”.
|
| 0:31:52 | DT | Yeah, we do the same thing. We provide the answer, but then show the user, well, this is the clause in context, right? You asked about this thing and here’s clause six, but clause six appears in the context of clause five and seven and you’re going to read the whole thing. We want to talk about guardrails. I think it’s at once a critically important part of developing an AI product for something where technical accuracy is important. And it certainly took us a lot of time to get comfortable that the guardrails we had developed were consistently applied. At the same time, it’s kind of, you’re inherently taking a bit of a risk, right? Because you’re leaving that risk management to an automated system. You’re sort of relying on the AI itself to identify inherently risky questions or inherently risky topics and stay away from them. So tell me a little bit about the process of testing and developing those safety guardrails, some of the things you learned.
|
| 0:32:51 | DW | Yeah, we started building our Lawpath AI around this time last year and then launched it around July. So we had a good four or five months to test and we did a number of processes. So one was we tested on our internal lawyers first, so we could kind of get as much feedback and a very fast feedback loop there. And then for the first three months of our AI system being live in production or live to the public, we had what we kind of call human in the loop or we actually had the answers being approved or denied by a legally trained person. So it wasn’t the most probably amazing job for that person, but they would wait for a client to come and use our system. And then when it did, it would trigger a notification to them to say, this is the question, this is the answer, is it correct? And we saw very quickly that at first they had to thoroughly review each answer and then it just got better and better. And each time we would find an incorrect answer, we would fix it, go back, change the prompt or whatever we were doing to fix it. And so the iteration was quite quick and got it to a point where we were pretty, or we were very happy with the answers it was giving. I do think though, we did build into our prompting or our instructions to the system that it shouldn’t ever, we almost put like an artificial line in there and said like, you’re not to advise on these certain areas, you’re not to advise past this level of detail. I think that will disappear over time, but at the moment that was sufficient. And look, I suppose we’re in a, maybe a lucky situation in that our clients are small businesses and often the type of legal matters that they’re tackling, the type of issues that they’re facing are not that complex. Some of them are very complex, but a lot of them are just your run of the mill issues. And so we found that for our use case, the level of kind of sophistication right now is enough. But I would say we pulled our data the other day. We’re finding roughly 20% of our clients are choosing to use AI. If they’re given the choice of a lawyer or our AI system, they’re actually choosing our AI system. 20%, which is, or which I thought was really interesting, especially when you’re given the choice of a lawyer, you’ve already paid, there’s no cost issue there, so…
|
| 0:35:19 | DT | Those are people who’ve already paid?
|
| DW | They’ve paid for the service. And the service comes with an on-demand lawyer or an on-demand AI system.
| |
| 0:35:28 | DT | What are the demographics of Lawpath customers? I mean, are they, I won’t make any assumptions, I guess. Tell me a little bit about the typical Lawpath kind of customer archetype. |
| 0:35:39 | DW | Typical Lawpath customer is small business in Australia, under 50 employees, typically under 10 million in revenue, and usually in their first 10 years of business. So to give you a really quick kind of real life of our client base, about 40% would be considered kind of startups and new businesses, a lot of kind of tech businesses in there in their first few years. Then we have a lot of kind of bricks and mortars, plumbers, design agencies, architects, those type. And then we have HR, accounting, professional services, so what I like to say, and I don’t want this to be an ad for Lawpath, but kind of if you’re large enough to have legal and compliance issues, but not large enough to have your own in-house lawyer or lawyer on retainer, then Lawpath is a great solution for you. That’s kind of where we sit.
|
| 0:36:28 | DT | Because that sort of opting for that instant digital AI solution, even when you have already paid for the human, it strikes me as kind of… maybe I’m wrong, but it feels like it would be a decision that would have quite a generational divide, in the sense that there might certainly be a class of customer who values the instant and convenient response of a software solution over the kind of personal, human, but perhaps delayed response of a real person.
|
| 0:37:06 | DW | Yeah, I mean we unfortunately don’t collect kind of demographic or age details, but I’m sure there’s something in that. I think, I was kind of digging into this the other night, because if it’s not a cost, if it’s not a cost factor, then it’s a speed or an interaction type preference, and so yes, if you want to speak to a lawyer on the Lawpath platform, you’ll probably have to wait between eight and 24 hours to get on the phone or get on a video, so there’s a little bit of a delay there. I think also maybe there’s just still this reluctance to get on a Zoom or get on a video call with a lawyer and explain your situation. People are so used to just chatting now. The other thing I did think was, are these a lot of the situations where they don’t want the full advice from a lawyer? They just want some kind of quick information. What’s the award rate? Should I be an operating and a holding company or just an operating company? Like these are the types of questions we’ll often get through, and maybe a full consultation with a lawyer isn’t warranted for those type of things. I think we also have to remember these small businesses, they’re busy running their own business, right? They don’t even have a spare half an hour to kind of, they want to create an employment agreement for the employees starting the next day at nine o’clock at night. They want to do it right then and there, and they can do that with all of their other services. I can jump on, pay my bills at 12 o’clock at night. I can do my accounting at 12 o’clock at night. I can do most things now through a platform self-serve at the time I want to do it, and I reserve my nine to five for making money. So I think maybe that’s an issue. I know our sales team provide a lot of feedback to me often saying that younger generations do just click to our model a little quicker than the old generations. So I mean, that can only be a good thing for businesses like ours that are really doubling down on this type of technology because everyone’s only getting older, so.
|
| 0:39:11 | DT | You know, the Productivity Commission and the Law Council of Australia did a bit of work on this when they were looking at unmet legal demand, that cost was only one of three sort of factors that really prevented people from going to see a lawyer. The others were that it was a really stressful experience. It felt kind of intimidating and like you might do the wrong thing if you go to see a lawyer, and also that it was just really confusing. how you access a lawyer, how do you find the right one. So, you know, it could be those kind of, you know, it’s stressful, it’s confusing, it’s not something I really feel like doing to talk to a lawyer, that might be driving people to pick the automated solution.
|
| 0:39:47 | DW | I think we, when we were running our marketplace in the early days, we always thought clients, especially in the SMB level, would choose the cheapest option. And what we actually started to see in the data that clients were choosing, we used to provide multiple quotes so you could get quotes from three or four lawyers and choose the one you wanted. And actually no one ever chose the lowest one. So cost wasn’t the main factor in deciding who to move forward with. It was often speed of service and the relationship that you built if there was confidence straight away. So, you know, one of the things that we put in place that got a lot of pushback was a review system. So every time you speak to a lawyer through the Lawpath platform, even today after every consult, you’d get, just like on Uber, a rating email that says rate them out of five stars. And lawyers used to complain about that all the time; “I shouldn’t be rated. You know, I’m from the legal profession, I shouldn’t be rated”. But actually what that did is it gave confidence to the next customer or client coming through. The peer-to-peer feedback, especially once you go online, you lose the trust face-to-face. So you need to replicate that trust element somewhere else and peer-to-peer reviews are the way to do that. And so if it’s not price and you can fix the trust issue, then it might be speed.
|
| 0:41:02 | DT | Yeah, absolutely. You mentioned that you had your own lawyers at Lawpath kind of dogfooding the early product for Lawpath AI before release. What was the kind of adoption journey like within Lawpath? Did you find that everyone was pretty receptive? Did you have some kind of hesitant users of the product? You know, I think you see some statistics published in places like Lawyers Weekly and the Law Society Journal about the extent to which our profession is enthusiastic about or willing to adopt this sort of AI copilot technology. What was your experience?
|
| 0:41:43 | DW | So I think I could summarize it as slow start, fast finish. And so what I mean by that was, yes, the lawyers, when we said, “hey look, here it is in staging, which is a kind of a test environment. Here it is, use it as much as possible. Use it with your clients. Provide as much feedback as we can”. A lot of pushback; “It’s not gonna be as good as me. What if it misses something? You know, I don’t wanna change the way I’m doing it”. All of those normal things that we hear when technology adoption comes in. But then, I mean luckily we’re in a position where we got to a point where we said, “well now you have to use it”. And I think once people started using it and people started realizing that there were efficiencies there. I mean, as with everything, it’s not perfect and it’s still not perfect. But there are definitely some solid use cases where it can make you more efficient and it can improve your job and your tasks. And then I think once they saw that, since then it’s been, you know, it’s widely used within the firm. We’ve gone ahead and some of the suggestions that came back from the lawyers. I think legal technology sometimes gets, we think about legal technology as it only creating efficiencies in legal work. But there’s also a lot of other applications within the business that you can use it for. For example, quoting or creating proposals. What a time suck. You speak to a client, you then have to create a proposal or a quote. This is what’s in, this is what’s out. You know, it can take a lot of time. There is a lot of proposal software out there. We now use AI to draft all of our proposals. So it already has a bunch of information or context about the customer. So it can provide a really comprehensive, detailed quote to them that feels really personalized. And we’ve seen an uptick in conversions on that as well. So I think we’re trying to use it in all aspects of the practice, not just for the drafting or the reviewing.
|
| 0:43:34 | DT | Yeah, nice. I think, you know, that brings out something that we’ve spoken about on the show before. When we talk about artificial intelligence now, we’re always talking about generative AI. We’re always talking about large language models. There are so many predictive artificial intelligence solutions, relatively simple things that I don’t think we would even describe as AI necessarily, just sort of machine learning techniques like regression models and categorization models that, you know, are largely untapped in law and making intelligent decisions about cost estimating is a great example of that. You know, we, most firms that have been operating for a substantial period of time have a wealth of data about how they estimate fees for particular classes of work. And they could be using relatively straightforward machine learning models to make better predictions about what the next matter is going to cost, both in terms of, you know, what are you gonna quote for it and what are you actually gonna record? Because those often vary. But, you know, we haven’t really done that. I think something that might come out of the enthusiasm for generative AI is an increased acceptance of predictive AI that is maybe less cutting edge but has just as much promise for a lot of those sort of business administration tasks.
|
| 0:44:54 | DW | Yeah, I think a lot, a big part of it is the profession just doesn’t have that much structured data. You know, those proposals are scattered all over the firm’s internet in all different spots. It’s actually quite difficult for you to get every single proposal, everything, quote, into one place where you can actually run some kind of analysis on it. But what I’ve found is that with the recent kind of hype around Gen AI and with the recent kind of ease of access, it’s got everyone starting to think more about where’s my data? It’s almost like being a catalyst for, well, to use this software and this software seems to be a game-changing piece of software for the industry. I need to line up all my ducks underneath to get it going. And so I think that not only will we see more Gen AI, but we’ll also see other AI, you know, machine learning, whatever it might be, have a high usage because it just can now be used.
|
| 0:45:55 | DT | I think you’re right. Everyone looks at these tools and says, “it would be great if I could do this sort of natural language interface over my own data. Hang on, my data’s really valuable actually. You know, I’ve got some really good proprietary stuff that I’m sitting on”. And then something clicks of “well, yeah, I actually hold something valuable in both the unstructured data that I hold, all the letters of advice I’ve written, all the contracts I’ve drafted, but yeah, also everything in the practice management system”. So it might be that kind of catalyst for thinking about our practices in a more data-driven way.
|
| 0:46:32 | DW | Yeah, exactly. And I don’t think it’ll be us as firms that need to make those innovations. It’ll be the practice management systems or the document management systems. They will all put a layer of AI on top of their systems that you can, I’m sure, buy for an extra cost.
|
| 0:46:47 | DT | As they’re all doing right now. I mean, who knows? They might not be things to buy. You know, we’re seeing both our sort of Australian native practice management system providers like Leap and some of their international competitors rushing to add large language model powered and generative AI solutions to their platforms. I wouldn’t be surprised if that kind of just becomes table stakes in the, you know, battle for practice management system market share.
|
| 0:47:14 | DW | Yeah, exactly right. I mean, I think it was 2023, it was just, everyone was just scrambling to see what they could get to market first. And then, you know, you had the big guys like Lexis building their own. You had, you know, the Thomson Reuters buying CaseText. So I think, yeah, you’re right. It’ll just become that’s actually just a sanitary feature. You know, you have to have some kind of AI. I do think that what we will see moving forward is that, you know, a lot of the use cases, it won’t be, “oh, this is an AI feature on top. It’s just AI’s in there and just being used and that’s just standard”. And so it’ll go from this, I have to pay extra or I have to do something different to use the AI feature to. I’m just using the software and there’s all, there’s something going on behind the scenes that I can’t see, which is an AI feature.
|
| 0:48:08 | DT | Ten or 12 years ago, you would have seen a lot of software solutions describe themselves as a “cloud X”, right? We never really see that anymore. We don’t think about whether something’s in the cloud. It just is, right? That’s sort of how software is done. I wonder if we’ll see the same kind of evolution with AI. Will we stop really thinking about the technology and start thinking about just the outcome that it produces?
|
| DW | Yeah.
| |
| DT | Our conversation today has really been about democratizing access to legal information, even democratizing access to legal services using AI. And, you know, we’ve talked a little bit about our own efforts in that regard, but we’ve also talked about how a lot of firms actually have really valuable data already and that a lot of these AI tools are actually, you know, relatively accessible. OpenAI’s GPT Store, for example, launched, I think, last month, which probably represents the most accessible kind of no-code version of creating your own AI chatbot to date. Do you think there’s scope for individual practices in Australia to contribute the democratization of legal information in that way?
| |
| 0:49:25 | DW | I do. I think that, I mean, this is a question that keeps coming up, which is, how deep do I go into this space? I mean, I do think that the large language models, you know, the big four or the big three will have however many they are, there’s no point competing at that level. My belief, and you may counter this, actually, there’s no point building your own model. You know, what I think, though, is the point of difference or what you can add as a firm with proprietary data is the database or the proprietary data that you feed into that model to then come out with a unique output. So I think that building your own LLM or building your own system is probably not the way to go, but I definitely think that finding an LLM that’s preset, maybe it’s ChatGPT, maybe it’s something, you know, as you said, a lot of the legal softwares are already building it in. So they’re doing all the hard work there, they’ll maintain it, and they’re probably using an Anthropic or a GPT or something like that anyway. And then you just can contribute to the system your unique value, which is your data or your know-how or your learnings. So I think that the great thing is even the smallest firms or sole practitioners already have access or will very shortly have access through probably a lot of the software they’ve already purchased to this kind of technology, and then it will be about just plugging it into your own systems. And hopefully it’s as simple as you’re already storing all your precedence in a system and there is, you know, and then you can just layer AI on top of that.
|
| 0:51:06 | DT | Yeah, I’m inclined to agree. I think, I guess thinking about the metaphor of cloud again, the infrastructure layer around highly performant, closed source, large language models, I think, yeah, a lot of the market power is going to accrue around some of those large incumbent players now, the OpenAIs, the Anthropics, Microsoft, Google, but that application layer, the use of the infrastructure for, I think especially domain-specific applications, and by that I mean things that are specific to law or specific to Australian law or specific to Australian corporate law. There is such a wealth of opportunities to take that kind of omni-use, highly general technology that is the large language model and tailor the application layer to something very specific. And I think that is something that even law firms can do. And it might be as simple as creating an AI chatbot for your website that describes the kind of services you offer and kind of gives that very high level explanation of the sorts of things you can do for a client. You know, we are a intellectual property practice, registering a trademark involves, you know, lodging these documents with IP Australia, it’s something we can help you with. So that’s a way to get past some of those barriers that we were talking about before that the Productivity Commission identified as real barriers to engaging a lawyer even for those who can afford it, being confused about the solution they need and a little bit stressed out about how to access it.
|
| 0:52:41 | DW | Yeah, I think that if you think of the providing services on a kind of linear line, if you can move that barrier further along so that you’re able to give some kind of value or service upfront at low cost, essentially to bring the client in and get them committed is a really good way to sell legal services. But to do that, we need to make sure that it doesn’t cost a lot at the front end. And I think there’s specifically AI, but specifically a lot of the software coming out now is really helping with that. So you can have a very low cost chatbot on your website that can summarise your services and chat to a customer. And instead of having two paralegals doing that, it’s a chatbot. I think a really good example, I mean, I’m a big fan of looking at software that’s outside the legal industry and then saying, how do we apply it to the legal industry? I really think that yes, we have a unique domain here in legal, but a lot of the time, instead of building stuff from scratch, we should just look to other industries and then just apply it within. And I’m a big fan of doing that. I mean, a lot of the time when we’re thinking about building a new product, we don’t start with the lawyers, we actually start with the developers at Lawpath because there’s a tonne of software that they’re aware of. And you start saying, “well, this is a problem I’m trying to solve. This is a legal problem I’m trying to solve. Oh, well, there’s a piece of software that almost does that over here that’s pre-built”. And then how do you apply it? So to give you a very quick example, Intercom is a piece of software that allows you to have chats with people on your website. And they’ve launched Intercom Fin, which is their AI bot. And you just plug it straight into your kind of internal document base of training documents. So we use Google Drive, all of our documents are in Google Drive. We just plugged it straight in. It took two seconds and suddenly the AI was trained on everything that we would normally train a lawyer or a sales associate. So that was a really cool practical way, again, of using it to fix that initial intake of clients. And I think we’ll just see more and more of that.
|
| 0:54:51 | DT | We’re nearly out of time. Before we finish up for today, I’ll ask you, I’ll ask all our guests, if you had one tip that you’d like our listeners to take away from our conversation today about democratizing the law, democratizing access to legal services, some of the opportunities that AI presents, maybe some of the risks, what would that one key takeaway be?
|
| 0:55:13 | DW | I think moving away, and so many people have already done this, but moving away from this concept of time-based billing to value-based billing just changes the entire incentives within a firm and the incentives are then to be more efficient and provide a better service using as many tools as possible because that always will lead to an uptick in revenues. You might see a short-term drop in client numbers or billable hours, but long-term you will always see a benefit. I suppose that’s kind of quite a niche takeaway, but I think that in order to democratize law, in order to provide a service, but most importantly, to fix that number you spoke about at the very beginning, which is 85% of clients globally can’t access legal services. Out as a profession, we need to think about that problem. We’re currently only servicing 15% of the clients we could service, and we’re doing a pretty good job. Legal’s a trillion dollar a year industry. It’s not broken, but it could be so much better, and I think one of the ways to do that is to make sure that we can provide more services to more people, and the way we do that is using tools and using software.
|
| 0:56:29 | DT | Yeah, absolutely. You know, I’m an opponent of magical thinking. You know, when I think about this 85% of unserved or underserved demand for legal services, I don’t think, “gee, wouldn’t it be great if there were some great large language model powered solution that could provide advice to all of those people, and the legal profession itself doesn’t grow at all”. I think the best way to improve access to justice is for every lawyer to be serving more clients at a lower cost, and the only way we do that is by, as you said, moving to a model that makes our services affordable, that’s going to have to be a value-based pricing model, and until we get rid of the incentive structures that discourage us from making our services more scalable, we never will, so we could probably spend another hour talking about that, but we’re out of time. Dom, thank you so much for joining me today on Hearsay.
|
| 0:57:25 | DW | Thank you for having me. It was really fun, thank you. |
| 0:57:28 | RD | As always, you’ve been listening to Hearsay The Legal Podcast. I’d like to thank our guest for coming on the show. If you’re an Australian legal practitioner, you can claim one continuing professional development point for listening to this episode. Whether an activity entitles you to claim a CPD unit is, as you well know, self-assessed but we suggest this episode entitles you to claim a Professional skills or Practice Management and Business skills unit. More information on claiming and tracking your points on Hearsay can be found on our website. Hearsay The Legal Podcast is brought to you by Lext Australia, a legal innovation company that makes the law easier to access and easier to practice, and that includes your CPD. Hearsay is recorded in Sydney on the lands of the Gadigal people of the Eora nation, and we would like to pay our respects to elders past and present. Thank you for listening and see you all on the next episode of Hearsay. |
You must be a subscriber to access this content.