He Left Cambridge, Walked Away from Law & Built an AI Lawyers Actually Use with Will McCartney

What happens when a Cambridge-educated humanities student walks away from a Sydney Law degree to build AI for one of the most risk-averse industries in the world?

Will McCartney did exactly that. At 23, with no law degree and no track record in legal practice, he started building Habeas - an Australian legal AI platform that finds real case law, cites it to paragraph level, and doesn't make things up.

Three years later it's winning Australian AI Awards, integrating with major legal platforms, and being used daily by some of Australia's most experienced litigators. This conversation is the honest version of what building that actually took.

🎓 The Origin Story

• He left Cambridge, transferred to Sydney Law, then walked away from the degree entirely - and the single administrative decision that made it inevitable
• He built a better product in crypto and still lost to vaporware - and nearly made the same mistake again with Habeas

🔧 Building Habeas

• The first version wasn't good enough - and he pitched it to a Big Six firm anyway. Here's what happened
• Why the lawyers who pressure-test Habeas hardest are the ones with 15 to 20 years experience - not juniors
• The moment compliance stopped being optional - and what forced that reckoning earlier than planned

🧠 The Hallucination Problem

• A lawyer used Claude to research a case and Copilot to check Claude's work. It still hallucinated. That submission went to Federal Court
• The hallucination problem nobody talks about - it's not fake cases, it's real cases that get twisted. And it's harder to fix
• Why Habeas searches first and thinks second - and why every legal AI that does it the other way round is gambling with someone's client
• Anyone promising zero hallucination in legal AI is either lying or doesn't understand how law works

⚖️ AI and the Legal Profession

• Big Law has innovation teams, enterprise budgets and dedicated IT. They're still losing to 20-person boutique firms on AI adoption. Why
• The billable hour is dying and most partners are pretending it isn't - what's actually replacing it
• 80% of Australian legal matters never get properly resolved. AI could change that. Or it could just make expensive lawyers faster. Which one is happening
• What he actually tells law grads who ask if they have a future in the profession

🏗️ The Moat & Distribution

• The technical advantage in legal AI is disappearing faster than anyone wants to admit - here's where the real defensibility actually lives
• What stops a well-resourced competitor building the same thing tomorrow - and whether that answer should make you nervous

💰 Bootstrapping vs Raising

• Money was available. He chose not to take it. What that cost him and what it gave back
• Why he's building a syndicate of lawyers who use the product instead of going to VC - and what that says about where Habeas is headed

🚀 The Future

• Why the next twelve months is about going deeper not wider - and what that looks like in practice
• Australia first. But the next markets are closer to ready than you'd think

If you're a lawyer, a founder, or just someone trying to understand where AI is taking one of the world's oldest professions - this is the episode.

🌐 habeas.ai 📧 will@habeas.ai 💼 linkedin.com/in/willmccartney

Transcript

SPEAKER_01 0:00

Welcome to the Ryan Tan Show. Today's guest is Will McCartney, the founder of Habeas, an Australian legal AI platform built around a deceptively simple promise. Make legal research faster without losing variability. Will's path here is anything but conventional. He began studying human, social, and political sciences at the University of Cambridge before COVID brought him back to Australia, where he transferred to an arts and law degree at the University of Sydney. Then in 2023, with a law degree within reach, he walked away entirely to go all in on habeas. The thesis was clear from day one. Australian lawyers needed AI that was grounded in Australian law, cited to a paragraph level, and designed for how lawyers actually work, not how AI demos look on stage. In under three years, habeas has won the Australian AI Award for the best use of agentic AI in the startup category, built integrations with Smokeball, partnered with ALPMA, attracted testimonials from some of Australia's most recognizable legal and legal tech figures, and is now used by litigators, boutique firms, and in-house teams across Australia. Will is 25 years old. Today we'll explore not just how he built habeas, but the harder questions underneath it, why hallucination is a catastrophic problem, specifically in legal contexts, and how you can actually solve it, what courts and regulators are saying about AI and litigation right now, whether the billable hour survives AI efficiency gains, and what a young founder sees about the future of legal practice that the incumbents are completely missing. So with all that said, welcome to the show, Will. G'day. Thank you for having me. And uh yeah, it's a pleasure.

SPEAKER_00 1:44

I really appreciate the invitation to the FSA. Yeah.

SPEAKER_01 1:47

Yeah, absolutely, Will. Yeah, I came across your product. I interviewed uh Damon Murdoch um as the first episode of season four, and um he plugged your product and you know had a little look into it. I'm really blown away, especially with you know AI in the space right now and people using AI for legal tech in Australia. So definitely want to dive into the product. So I wanted to start off with your origin story. So you studied at Cambridge, you had a bit of a passion around politics, sociology, and social sciences, but most people wouldn't connect that straight to AI. So what actually interested you uh back then?

SPEAKER_00 2:21

Yeah, you know, what interested me back then and probably still continues to interest me is, you know, the broader discipline of the humanities, right? So even in high school, I was quite interested in English history. Um, you know, in my HSC year, for example, I wasn't, I wasn't terrible at maths, but I decided not to take maths just because it was it was not really my area of passion. Um and yeah, I think I've always had a sort of deeper fascination there. That's where my skill set kind of lay. Um, so it was a natural progression to try and apply to Cambridge in uh a culture that is probably a little bit more uh rewarding and kind of they prioritize the humanities a little bit more compared to the way we looked at it maybe in Australia sometimes and just from a from a university level. So I did like the fact that Cambridge had this sort of wide-reaching degree where you didn't need to specialize in just politics to begin with. You could look at sociology, politics, history, IR. And yeah, like I one thing that that even that first year at Cambridge did teach me was like sociology in particular, was an area of interest um which we hadn't really ever dived into in a school context. So, you know, it's sort of my first time studying just like the some of the more classic sociologists like Max Baber and people like that, and I'm like being familiarized with their theories. Um, but yeah, I think more broadly that um always probably ties in that background that I have to any of the work that I'm doing, because it influences kind of just the way that you're you build products for people and that you um understand what people want, how they're viewing the world, um, the sort of risks they're willing to take, all those sorts of things. So there probably is always just these connections. And then second, secondly, it's we're now in a world where like very rapidly technically things have changed. Um so some of my coding skills that I developed in 2023, 2024 have been completely um, there's no competitive advantage there almost in any sense, in terms of technically um AI is getting better and better on the coding and on orchestration front and a lot of different technical skill sets that we traditionally thought were just like human domains. Um, but I do think that broader humanities backing is is is useful still for anyone in terms of like how you can communicate an ideal, even in a business context, maybe speak to a customer or something like that has a lot of like micro ramifications, right? Um so yeah, I still think it ties in, even if it wasn't the obvio you know, the obvious move to go from where I was at to where I'm now.

SPEAKER_01 4:54

Yeah, totally. Yeah, you've you've described yourself as um studying, you know, up to four units of history in the HSE. You know, you're a passionate historian, but you're also passionate about ethics and it kind of ties in, and it gives you a really nice background to pivot into what habeas is today. So I guess COVID brought you back to Australia and law school. What was that transition like? You know, did you feel like you were starting over when you came back?

SPEAKER_00 5:16

Yeah, I think it's a great question. I think uh in a way, it was kind of like everyone, I don't think it was just me around that sort of period of time had different challenges, whether in a health context or a family context or um, you know, international like students like myself, it definitely throws a spanner in the works. Um, and you know, kind of the most you can do in that period of time is to just kind of reassess and think about, okay, well, yeah, it does feel a little bit constraining. Like you, you chose one path and this quite unpredictable world event um kind of occurs and you know impacts you and your family in different ways, sure. But um, you know, skeering out of that, I do even think, you know, there's some positives to really be taken just from a perspective of if I if that hadn't happened, I might have just kind of gone down a more traditional career path. And, you know, many students out of Cambridge, they might move into something like consulting or fields like that in their first few years after you need to experiment. And it wouldn't have even necessarily been a worth, a worse bad path. It's hard to say, right? You can't predict. But I do think just the fact that it allowed for this period of time where people were freely experimenting with different things, like in their bedroom, in you know, when things were locked down in particular. And even in those that first year at UCID, there was still like very strict kind of um regulations on kind of like movement and what you could do and who you know, who you could speak to and where. Um, and so even during that time, like I was actually creating a bit of music. So I was doing stuff that was just exploring passions, interests, stuff that was completely external to a university or academic context. And I still play a bit of guitar and do stuff like that. And I I do think there's a link between anything, you know, creative that you might be doing and sort of entrepreneurship more broadly, right? Um, so yeah, I think suffice to say it, you know, a bit of a a tough time for everyone, but also something that leads you in a different path that you wouldn't have necessarily chosen and brings new opportunities as well.

SPEAKER_01 7:19

Yeah, totally. And having that creativity, as you mentioned, it's a huge hill. I know you've mentioned yourself being a bit of a techie, you know, building computers and gaming, and that kind of works really well together with, like you said, the music side of things, but then also the broader political sciences and ethics side of things. Now, I guess at which point of the law degree did you kind of see the fork in the road and realize it wasn't the destination?

SPEAKER_00 7:43

Yeah, I think the situation again at that time was complex. It was more that I was studying, but then I was also increasingly interested in technology. I was doing things on the site. Um, so even if it wasn't necessarily anything that huge or serious, um, I found like this interesting division where I was going home uh from university, doing my study, etc. But any time I could allocate to things with a sort of more self-learning angle, you know, I would. So even kind of um, you know, after the first year or two, I can't remember the exact timelines, but yeah, I started finding that I was um doing more and more of that in a bit of a schism or a split in my attention. And also just that I found that, yeah, in a humanities context in particular, maybe um I felt like confident enough that I was like, okay, I kind of already know how to write an essay and do some of these core skills, which I've actually been almost doing since I was a kid to some degree, and like learning all the way through high school and early uni. Um, and it was just kind of that sense of when there's probably a bit more fun, but also you're just learning more in a different context, which is like self-learning, doing a bit of early entrepreneurial stuff. Um, yeah, like it just leads to a bit of a changing of priorities, perhaps. And um, further to that, there was also some struggles around like with you City in particular and actually just getting credits for my previous study in um Cambridge, which was unfortunate, but that also provided this sort of like there was a few external push factors as well as internal pull factors, you could say, at that time. So yeah, I don't actually necessarily think about that stuff too much anymore. It's kind of just um, you know, you you work with the cards that you're dealt at that particular time. Um, but yeah, I think when you're someone who's just like kind of more obsessed with learning, and then I had that bit of that pull towards doing my own work, um, you know, my own projects and side projects, and then that sort of there's a point later on where you actually make a firmer decision around, depending on what's happening, whether to say defer your degree as well. Um, and I also want to clarify it wasn't kind of like, oh, I'm dropping out of uni. It was more like that first year of habeas, we were building, I was building it on the side while studying. And then it was kind of like um further to that in 2024 when I saw there was more validated opportunities and the scale of the market, but also the time required to execute on it properly. That was kind of when I was able to make a more like data-informed decision rather than just saying, oh, I'm just gonna drop out, which wasn't the way I was looking at it. I was looking at it as I can defer my degree for the time being, allocate my full attention to this. And if it turns out to be the wrong decision or just I miscalculated here, you know, uh it you can go back, you can repursue, or you can yeah. So it wasn't like an either-or drastic sort of decision necessarily, yeah.

SPEAKER_01 10:39

Yeah, it's fantastic. And and like you said, it doesn't have to be a binary yes or no. You can then defer for a year. And what was going on at the time with COVID gave everyone a bit of time to kind of think about other projects. Uh, were your family supportive of the move? How do they take it when obviously you um, you know, you you told them you wanted to kind of as defer and then more later on just kind of pursue habeas full-time?

SPEAKER_00 11:01

Yeah, no, I think I've always had um great support, like from a from a family context in a lot of things I'm doing. Um, and yeah, we we took a look at you know the situation. I've I always think that um it's great to consult with family and people in your network beyond that as well on important decisions. So um, yeah, it was definitely something there was there was alignment of understanding what I was doing um and and why, and also the fact that you know it wasn't, as I you know mentioned before, a final decision. I I think that just applies to a lot of things in life in general. So you often, when you're in that moment, you feel a great, you can feel a great sense of tension or fear or uncertainty. Um, but you often realize, especially when you talk to people who are a bit further on in their careers, who have completely, you know, even reskilled, taken a totally different degree. Um, they've gone from law to medicine or something like that, and invested that time and degree even later on where there might be a perception of some cost. But really, you know, most points in life, you're never necessarily too far behind to make a pivot or a different decision as well. So I never look at anything as being too inherently final um when it comes to decision making. Um, and certainly at the time, yeah, it wasn't like, okay, it was like there's some ways you can mitigate this so that even if there was something terrible, like, you know, the market crashed or there was a global financial crisis or something that you couldn't predict and you had to go back to a more traditional uh career path, you know, I would have been ready for that as well. Um so yeah, ultimately I think, yeah, always a great idea to consult with people your network, but also yeah, family can be an extremely strong support as well. And I also think further to that, actually, um, sometimes family are the people who can be early believers who know you better than the people you might be then going and pitching to in the market who don't know you at all, right? The family might be the people who know um what your skill set is and and and what you should focus on and yeah, whether or not it's the right opportunity. So yeah, definitely hope. And on a personal level also, yeah. Yeah, my parents are awesome. So yeah.

SPEAKER_01 13:10

Yeah, it's awesome to have that support from family and and the backing, especially in the early days as well. So um look, that's great to hear. I guess moving on to building habeas, you know, walk me through day one. What did you actually have? Was it an idea, a prototype, or did you have a paying customer or even a pilot?

SPEAKER_00 13:28

Yeah, definitely no pilots, no paid customers on day one. So um I think when I started out, I was ideating about around a bunch of ideas just in the AI space more broadly, um, trying to grapple with like where is this actually going to go, which fields are gonna be heavily impacted and which aren't. Um, and also, you know, yeah, you're kind of making these predictions broadly around if you want to allocate your time to building something as well, uh, without getting too lost in the weeds, right? Of you know, overthinking it. Um, but yeah, I think at the time it was more actually just born out of experimentation. I was building a few separate prototypes for a few different ideas. And again, mostly in the first few months of even just starting to work on what would later become habeas. Um, I was actually looking more at the idea of like seeking a technical co-founder. Um, so I was interested in more kind of seeing, look, who in my network, or is anyone I know who'd who'd be really interested in this? I think in my personal network as a student, not really, because most people just sort of focused on uni, um, wasn't top priority, wasn't something they were looking at as a serious pursuit. Um, I did manage to meet someone who was, you know, really excellent from um, you know, as a sort of part-time engineer who was interested in the project from actually the CSIRO and had a really great machine learning skill set, which I didn't have, but um, that probably wasn't the right fit because of his full-time commitment to that job and um the time that would actually be available to invest. So I think from that first couple of months where I was ideating and then also seeking a technical co-founder, there was also a point where I decided, look, I'm just gonna have to upskill in some of these areas. But also being aware that there's like a bit of an opportunity there because the AI stack itself was still emerging. I think even now in 2026 still is and will continue to be. It's it's not like uh there's one agreed upon way to build um AI products. There's firmer frameworks in place. But even then, I could kind of say, well, look, the advantage here is that um it's a newer stack. So you can kind of learn as you go, be quite deeply embedded in it to learn how to build a particular type of product with a very specific architecture. Um, and so yeah, it ended up being this situation where I was like really focused on just developing my own skill set there, which was it which was a journey. Um and I think that had implications for like in 2023 in particular, my front-end skills were rough, which for anyone who's from a non-technical background is sort of like how the app looks and and feels to the end user. And the initial prototype that we kind of got to by maybe sort of, I think roughly maybe August of that year in 2023, um, was just like a really simple first draft of where the product would go, but it had the the basic bones in it, right? Just this idea of citation-backed responses and when someone typed a query and wasn't just broadly hallucinating, it was drawing on real data. So it had the very initial bones, but it wasn't able to handle more complex legal queries, more multifaceted legal queries at that time. And so that was kind of like getting that prototype ready was in particular, because we had a trial with Allen's at the time, which was exciting. But um, again, tech was all still so early stage. And I think as well as a, you know, someone who is still broadly non-technical in terms of my initial background and became technical over time, um, yeah, like I saw that initial pilot back in the day as um yeah, a learning experience as well, because the tech wasn't quite yet there broadly across the mark and it wasn't there yet for habeas. Um so yeah, I could go into that in a whole lot more depth, but kind of just when you start out, you're moving from um initial idea stage, thinking about what core problem you want to solve. Um, broadly for me, that was legal research and problems associated with it, because the legal research stack has always been a bit clunk clunky prior to AI's emergence, I think, and um tricky to navigate, especially traditional search databases. So it was kind of uh and and still is an ongoing journey in terms of refining a product and finding product market fit and stuff like that. Yeah.

SPEAKER_01 17:55

Yeah, fantastic. We'll go a bit more deeper into that. I just want to understand, I guess, how habeas works in a future section. I guess, you know, when you ran the crypto startup before habeas, you know, you mentioned you lost out to vaporware products with no substance that were better marketed. Did that kind of impact or influence this habeas journey and especially those early stages between balancing, say, product with say sales and marketing?

SPEAKER_00 18:18

It did impact the journey, and I think in some ways in a very positive way, and in some ways in a less positive way. Um, I probably actually should have taken some more lessons from that. I assume that like when I first started and built a sort of project in um that field, which was like um, you know, we were sort of at the time trying to build a game for the blockchain. And my thinking was kind of, oh, look, there's a whole bunch of products out there that are quite clearly either trying to just they don't have a product, they're raising a lot of funds for something that doesn't exist. Um, you know, my rational brain sort of went, okay, let's build something real that leverages some of the most interesting um, you know, mechanisms that are enabled through blockchain technology and stuff like that. Um and yeah, definitely a bit of a lesson at the time was no, a lot of the projects that were successful were just purely in that market, you know, appealing to almost the betting mentality of people and the way because it was such a new tech, people just kind of like looking to hop into different, you know, even different tokens and different chains and stuff and make money. And um, I guess my initial assumption was kind of like as I moved on to more serious business technologies, I was kind of thinking, look, I'll leave that space behind because it's not mature enough. Um, and it doesn't necessarily have the death. But I think probably what I could have learned there actually is just still marketing and distribution always kind of reigns supreme, even in business context. I probably should have taken that lesson even more and realized that in legal tech, some of the same dynamics actually in here. Um, but the positive lesson from it was just this. Okay, well, I also want to enable myself to have a broader understanding of the technical stack because when I was uh building that previous project, we were sort of hiring a couple of developers to help with the blockchain side of things. And while that's that was great, um, yeah, I wanted to empower myself to know that whatever I was building, I had a hand on the tiller, I knew was almost building that stack up myself, and I was kind of developing my own technical skill set um further to that, right? And yeah, I think that that being a positive because now even as habeas scales a bit and we have some help um in functions of sales and growth, and then probably you know, in the next few months as well, engineering, but it does help if you have that score skill set yourself, at least to some degree, and then knowing who to work with and things like that, right? So yeah, I think in in some ways, sometimes you go through experiences, you learn some lessons, and sometimes you even like slightly learn the the the wrong lesson and you see certain patterns popping up as well. Um and you're like, oh, that was the lesson I maybe actually should have learned and been aware of. But you know, that's all good and yeah, it all contributes to your growth when you're in your sort of early 20s.

SPEAKER_01 21:19

Yeah, totally. Yeah, it's a personal evolution, and I guess to bring it back to the marketing sales construct, um, you know, you've you've done a good job. I mean, you've got a great landing page, you've got demos on there, you've made it easy to understand. I guess there's always a bit of tension, right? Because you want to focus on the product development and make it as good as you can, but there's always that fine line between, you know, not getting too deep into it, but making it something that is in a marketable and kind of prioritizing those skill sets. So look, that's fantastic. How did you get your first law firm to trust a 23 year old at the time with no law degree and no track record in legal practice?

SPEAKER_00 21:54

Yeah, I like that framing. Yeah, I I think it was more in the sense that firstly When I was pitching the product to firms, it was something new, something they at least wanted to, especially because my strategy at the time was probably talking with some larger firms, actually. Um, right or wrong, it was the idea of look, okay, maybe we can get incubate the product somewhere, get a cornerstone client, perhaps, or something like that. Um, yeah, I think it was actually an ongoing challenge, to be honest. And it's not something that you can easily mitigate because a lot of large firms and in particular in enterprise settings, they might be more reticent towards even a good or promising product that still feels early stage or still feels like um, in some sense, it's nascent, right? Uh, so you know, yeah, there was actually some challenges along the way, 2023 to 2025, really combating that and even sometimes getting your foot in the door with firms. And one of the things we're now at in 2026, which is kind of different, is that there's almost this um people try the product and kind of kind of see for themselves and can even be sometimes referred in by someone else who's used it and compared it to other products and had good things to say. So we kind of sidestep that issue a lot more because it's kind of like if someone comes at me with, oh, but I'm not sure about your experience, we haven't a weight of evidence and proof that I didn't have at that point in time. So yeah, it's kind of a slightly different situation. But for the firms that were, I think, interested, it was more just framing it as, look, let's run pilots, let's allow your lawyers to get some initial an initial vantage point, particularly late 2023 to mid-2024, a couple of the bigger firms we piloted with. Um, at least give them a taste of this. And um, again, there were some mistakes I made in terms of maybe how I structured some of those things. But we managed to, you know, I guess we managed to break through because the prom the product itself showed product promise. It was a space that law firms were becoming more aware of, even though they were still reticent to, I think, adopt and fully best in anything. Um, and you know, sometimes when you have something that at that point in time is fundamentally new and you can explain the rationale for you know where the future is going and why that product might continue to get better as well. So also I think an enterprise vi setting, you should always be looking at things like vision as well, um, and not just uh saying where the product is at now is as good as it's gonna be, right? So yeah, that's what allowed us to get through and build some of those relationships. And even if like at the time there was like some of the larger firms in Australia that, you know, we had uh pretty in-depth uh discussions with, or even like the pilot with Allen's back in 2023, which was very much, you know, it was agreed on upon to be a provisional pilot. It w there was no guarantee of, you know, look, if if this is successful or something for move forward, it was more like a way for them to just get some insights, deliver some feedback on the product. And I think um the knowledge manager at the time who who structured that pilot was like really professional and excellent, excellent in terms of we got heaps of data and feedback from it, like full-on graphs and um like responses from users who are like, okay, you know, like this is a seven out of ten, this is a six out of ten, here's what it needs to be able to do for me to be fully happy with it. And I think you know, that in particular informed a lot of development. And then simultaneously, when we did some other pilots in 2024, um, that informed some elements and development for that year. Um, so yeah, like ultimately, I guess in answer to the question initially, it's just it's a it's a trust building exercise over time. And also probably early on, I could have gone with a slightly different strategy as well in terms of who we were working with. Um, yeah, but that's okay. It's just a learning process over time.

SPEAKER_01 25:56

I'm just having a look at your website here. So, you know, you've got some great names here. You know, you've got general counsels, principals of law firms, um, startup founders, you know, litigators, I can see, legal counsel at UTS, special counsel, and barristers, um, I think separately. You've described yourself servicing lawyers with 15 to 20 years, PQE. How is that? Is that that's probably a very different conversation when you're speaking with you know these smaller boutique firms who are actually showing interest and love the product.

SPEAKER_00 26:26

Um yeah, for sure. And and I really think again to emphasize here, there was a lot of probably when you're again in the more nascent stages of the technology, you you have a vision of where you think as well the product will go. And then there's always a slight gap between where it is right now. And then you've also got um in the legal field, in particular, selling to lawyers, compared to some other spaces that you know have friends who are founders in, um, the bar is exceptionally high, right? So you've got you've kind of got these extra additional standards and the importance of like the faithfulness of the answers that you're giving in a in a legal context, you kind of need to be watching out for. Um, how intuitive the technology is to an everyday lawyer, all these kinds of different factors. And so I think it's, you know, it's been a bit of a journey where, in particular, early 2025 was when we started getting like some real self-serve customers on the platform because we actually kind of launched, launched a self-serve dimension, which is where some people could just kind of sign up themselves. And in particular, principals of small firms, sole practitioners, etc. And so that became a customer base that grew outwards. Um, but particularly for more tech-savvy lawyers who really wanted to dive into AI and kind of understood some of the limitations, but also some of the strengths of a product like habeas compared to, say, just using Chat GPT. Um, yeah, we got some early believers in that year in particular that I think um, even where we hadn't necessarily quite broken through on a few enterprise opportunities, um, it was good to build out this base of sole practitioners, smaller firms, and even over time, you know, a couple of barristers and people who've come on board, and now in particular in 2026, that's like a market where interested in it. So, you know, then you have this second dimension of like when lawyers are many years qualified, they also know the law in that space really well. So they can kind of draw your attention to issues in the product at that time, or could it be more capable of doing this? You just gain a whole bunch more insights, feedback, um, thoughts on how to refine certain things that aren't available in the product, things that you just haven't engineered quite in the right way to make sense in a legal context, all that sort of stuff. So I do even think though, in a way, sometimes it's this dynamic of the bar for a soul crack who's specialized in a field and what licenses they'll take on is sometimes even higher than sometimes with some firms where they actually might go the opposite direction and they might start out with like a pilot of co-pilot or just something that they get through their Microsoft subscription. And um that's led to some interesting dynamics where some like mid-sized firms in particular have kind of been a bit burnt out or fatigued with AI. So they didn't necessarily go to a spot, a specialist or a boutique product like habeas initially. They might have gone and tried copilot and said, Oh, this AI isn't good enough for legal research. I need to do Australian specific stuff. We're just going to put this AI thing on the back burner, right? Soul practitioners do have this advantage, and even just like small firms like that are especially more nimble and innovation-savvy people, like you know, Damon Murdoch, who you previously interviewed, um, they will just go and try a lot of things. Because it for them, that's just let me allocate a couple of days when I'm not got any active briefs to just trying different tech, researching the market, doing a deep dive of what's out there. They're probably often more likely to engage with like newer products that might also be higher quality, even if they don't have an associated massive brand name behind them. Um, so yeah, I think there's lots of positives as well to working with people who are in sole practitioner settings or in-house counsel now increasingly who are interested in habeas who might just be like a couple of uh general counsel and you know, sometimes they still are massive are managing massive risk and massive matters, but it's just a couple of people who are subscribing, right? Um, so yeah, hopefully that kind of helps answer, you know, the idea. There's so many other ideas, I guess I could cover there. But um basically it's just been a really good, you know, to have some of these experienced lawyers um holding us to account and thinking about what else could be improved and you know, especially early stages, when when did the product fall down and how could that be refined?

SPEAKER_01 30:52

Yeah. Yeah, totally. No, that's really cool to see. And like you said, you know, the the data itself is interesting and it shows you what they're looking for in terms of how to structure the product and to use it to kind of better it, but potentially around product features. I know you've made a bunch of tweaks and updates that you publish online, which is cool to see. I guess just to take a step back for the audience, what does habeas actually do in plain English? Let's say I'm a lawyer, what does my day look like with it versus without it?

SPEAKER_00 31:21

Yeah, okay. I mean, that's good framing. So I think I'll qualify this by saying different people in different practice areas probably extract different kinds of value from the platform, right? But a at a base level, habeas is a way where lawyers can ask natural language questions for quite complex either situational legal queries that explain, you know, maybe a situation a client is going through. They could also use that for case law research, like retreating authorities legislation. Um, and what habeas is really useful for is not only extracting information from a bunch of sourced material, but compiling that information into something usable, like an initial research memo, a first draft or something that can be um worked upon later. And so we kind of traverse this field from um, you know, someone's got some a complex issue they need to get a first pass across. And habeas can help with help them with that. And then we do a whole bunch of probably sub-functions within that that are valuable to people. So some of that might be document analysis where people can upload their own documents to the platform and use habeas for reviewing and analyzing briefs and kind of getting this second pair of eyes for strategic functions, um, like analyzing, even down to, you know, what is opposing counsel said here, what's the validity of their argument, is there any weaknesses? Um, they can also use it just for pure sort of case law search, just kind of trying to find key authorities, key information, and down to the sub-paragraph so they can always see the sources. A lot of people might use it because they want to mitigate the hallucination risks that can appear with using traditional AI as well. So they want everything to be um showing a real source, showing a real court website and where that information tracks back to. Um and then there's increasingly functions in 2026 as well for really sophisticated drafting workflows. So while we always, I think, are building from this basis of research and search, um, acknowledging that a lot of people started using habeas for drafting, even though that wasn't necessarily core to our initial vision, people want to just kind of do a lot of things in the same platform. And so increasingly we've introduced features which allow them to um magically draft via an existing template structure and take information that they might have ideated on, researched, added additional context to habeas in a chat, and then um turn that into a first stage work product. Um, like that could just be again a research memo or or a brief, or um it could be something like um within an LOA structure that they've uploaded for the platform, which is specific to their firm, because you know a lot of firms will have precedents or set ways that they do things. Um, so bring those workflows together all in the one platform. So then to dive a bit deeper as well, um for someone like a barrister, um, their main utility might initially be just getting across a matter and the area of law in depth, because they have to be across so many different areas and so many different matters, and they've got lots going on, and that's really useful for them. Um, but even similarly, someone who's like in an employment context and they're really that's the only practice area they specialize in, they might use habeas for researching particular interactions between clauses in contracts and say fair work commission guidelines or um how that clause has been applied in far in past case law. Um so again, you could outline that for every different practice area, and where someone dives in and what they really want to do is a little bit different and where they can probably extract the key value.

SPEAKER_01 35:09

Yeah, totally. Yeah, I've just opened up my account here on habeas. You know, I've got a trial in preparation for the interview. Yeah. And yeah, it's really good. I mean, I I love it that you know you've separated it into the search engine, you've got the assistant there, um, documents, so that's great. And then you've got precedence as well, as well as the quick chat, which is cool. So, you know, I like how you've got not only documents, but you've got full litigation prep assistant, employment law assistance, goes down to, you know, family law, IP, insolvency and restructuring, corporate governance and regulatory compliance as well. And I like how you can kind of put the documents in, but when you put queries into habeas, what is great about it is that it will pull data, firstly, as you mentioned, case law and OSLE judgments and stuff like that, but it refines it in a way that a lawyer would think, which I think is important when you use, you know, let's say just Chat GBT or Claude, it might introduce a lot more narrative, which is not necessarily bad, but I think in terms of the usability and and framing for pleadings and things like that, habeas really structures it in a tight legal net.

SPEAKER_00 36:15

Yeah, 100%. And we give people, um, especially those who are willing to experiment on the platform, we give them those different modes. So, yeah, again, you've got the quick chat, which is designed to give um shorter, sharp synthesis style um uh answers by default. Um, but at the same time, how someone prompts still matters in habeas. So if they ask for, we've kind of tried to make it sort of if they ask for a deeper level of depth or a deep dive, habeas will pick up on that in the answers that it gives, the level of sophistication that it sort of then imbues into a final structure for a structured answer to the user's question. And yeah, some people will probably, like quite a lot of people probably don't experiment with some of those deeper customizable features, right? But for those that do and are really like that's kind of our our market and audience, they can yield some really great results. And I think this is where with Legal AI in particular, it comes down to more nuanced things, which you know involves, for example, the tone, you know, the structure of output, the voice. To what extent does the assistant that someone's using feel like a thought partner rather than just something that's just parroting back certain things they've already said? So one issue with ChatGPT, for example, is that in its default system prompt, it can be very sort of affirming to the user, just affirm whatever they say, almost to a point of obsequiousness, or just sort of saying, yeah, that's right, or you know, yes, definitely. With our structures, we want more pushback, more sort of context-informed material, something where it feels like to the lawyer, like it's a bit of a thought partner. And it also, you know, some of our GCs have commented on like the character as well being different, which is something we try and imbue in our system design all the way from top to bottom. So again, kind of the character of what it is and how deeply it kind of understands just off a first query what it is you're really trying to get at and whether it can go do that or not. And of course, we're not always gonna hit the mark. Sometimes habeas will go down a rabbit hole, especially if someone just gives a really simple prompt with not much context or explanation. Habeas is gonna have obviously less less grounding around like where to take that person. But what we see is that, you know, people I think are increasingly understanding they can do some more deep dive prompts or like ask habeas to situationally break down, okay, you know, I'm a plaintiff lawyer. I really want to look at quantum for miso, dealyoma, case law. And um, I want to extract any useful insights, build me up a narrative of what's happened in the past and why, and then apply that secondly, in a second prompt, they might say apply that to my current matter, just as a strategic function, kind of what would be um the likely sort of guidelines here based on past precedent. And of course, you know, then they're bringing in their own experience, their own deep understanding of the client and what they've seen in practice as well. So it's not just like sort of replacing that because plenty of times habeas might show that oh, the the the precedent says this, but of course, strategically someone say, Well, no, we can argue for a new foundation here, or because of the specific nature of this client's case, we know that we can hit a higher bar in that in that scenario, for example. So yeah, still very much a very deep element of skeering that we built in. And um, a lot of times people might get to a first draft, first research memos and compile that elsewhere, or edit that draft from Microsoft Word or whatever, however they want to go about it as well. Yeah.

SPEAKER_01 39:52

Yeah, totally. And when I interviewed Damon, you know, he shared a case that a lawyer had used AI. I think he had gotten thrown out of court because the AI had hallucinated and gave a bunch of case law which didn't even exist. So I think that's definitely the hallucination problem has been a huge thing, especially using AI in a legal context. I guess for people who don't know, what is AI hallucination and why does it happen?

SPEAKER_00 40:17

Yeah, I mean, I think at this point a lot of people are quite up to date, at least in legal, in this idea of I think firstly with hallucinations at a very simple level, it's sort of when um a generative model outputs something that isn't factually accurate, either it involves misinterpretation or sort of misapplication, but more broadly, the more egregious cases are when something is just um flat-out untrue, like a case that doesn't exist, um, that's just been j generated. And um, you know, those are the more, you know, probably the riskiest and the things that have gotten people in in deep water, right? Um, and really it's more just like a function of large language models themselves. So to it, to an extent, it's something that, first of all, we should clarify and say that a lot of big labs and and people who are building LMs are deeply aware of this, and it's something that models are getting better and better over time in terms of how they reflect truth and accuracy. But it's also something inherent to how a large language model works with like next token prediction. And so without getting too technical at the moment, that's kind of almost like a risk you need to mitigate. And there's a few ways you can look at doing this from an architectural standpoint. Um, one is you might have a very search-first um and inference later style system, which is something that habeas does. So it will actually really be biased towards searching information. The summary on top is more like a compilation of material. And it's it's always kind of taking those initial search steps first, unless a user sort of requests the opposite for a particular reason. There are also innovations around, you know, people who might look at other types of AI models like encoder models and uh that have been experimented with in the law that might allow us to mitigate hallucinations or don't have those same kinds of errors. So basically, main thing I want to emphasize to people who may or may not be from a tech technical background is there's heaps of work being done on it. And there's already systems like habeas that kind of greatly, like, you know, as someone like Damon would utilize it, allow for that observability of where information is coming from. Um and I think also then there's yeah, entirely new AI and machine learning techniques that are emerging and different systems that people are experimenting with. Like the team I think I've seen at iZachers in Australia is doing some work on this and you know, seeing if they can solve it through it through different methods. But yeah, I think this all alludes more broadly to the the conversation around hallucinations where there is a real risk, but also in Australia we have a tendency to only focus on a risk. And I think that's something that set our AI landscape probably in legal a little bit behind because what we've done is we've kind of spent, you know, quite a bit of 2024 and 2025 just highlighting a lot of like, oh, this is happening, rather than pointing to solutions or kind of like, oh, there's platforms like habeas that we can use that enable us to, you know, ensure we're sourcing real case or and real data and have a search first, inference second like approach, right? Or you know, thinking about how can we as an economy or you know, the government broadly um invest in initiatives or whatever it might be to kind of um help solve these different issues. Um so I think, yeah, more broadly, it's it's one of these things models themselves at a base level are getting better, best, better on the frontier. And then secondly, that's why we build in more complicated architectures under the hood. Um, that's not to say that there aren't plenty of products that still experience hallucination issues, which to me is a shame that some of them have just been adopted so wholesale into the profession. I think It's a real risk with some of the products that have come from from overseas that are still exhibiting hallucinations, and people are sort of quite egregious ones as well, where just again a case doesn't exist. And so that's kind of a separate issue, but which is it does matter what stack is underpinning illegal AI's um whole architecture and the results you'll get and the quality of results. Yeah.

SPEAKER_01 44:23

Yeah, totally. No, I I mean it makes a lot of sense. And I guess you've got two kind of issues of hallucinations. One is the ones that you can see that are clearly wrong. The second ones are the ones that you don't know exist. And obviously the audience and people using habeas are quite switched on and they know kind of what to look for and they're using it as a complementary tool. But I guess you know the independent benchmarking of commercial legal AI tools has found non-trivial hallucination, even in systems that are built for that purpose. Do you have a target for Australian law queries in terms of mitigating hallucinations and how do you measure it, especially those ones that are not obvious?

SPEAKER_00 45:00

Yeah, I think there's a few things to say here. So obviously the target over time is kind of um broadly, you can say zero, but it's actually probably more nuanced than that. And I and I always see I think it's a moving bar. Um to clarify that, first of all, the law itself is shrouded in a lot of subjectivity. Um, unlike a discipline where mass uh like mathematics or certain elements of science where we can be pretty objective in terms of like setting up a formula and saying, you know, A equals A or whatever it may be, um, the law isn't necessarily like that. So there are some objective things that, of course, exist in the law, but there's also this element to which the law is contested, it's moving, it's dynamic, it's subjective, decisions can be overturned, right? Um, so we actually want to be allowing for that fluidity to a degree in any platform rather than than trying to prescribe some kind of um almost mathematical certainty at a hundred percent layer, where it kind of becomes more important is just ensuring that yeah, in habeas' ability to say analyze the interactions between case law and legislation, the lawyer is equipped to use the tool as well as possible so that habeas has insight and oversight over the right information relevant to that matter, that practice area. And at the moment, I still think that's slightly a skill set thing. Um, over time, though, habeas will be become better and better at intuiting what the lawyer needs and how to qualify any answer it's given as well. Um, so yeah, I think broadly it's it's a quite complex question, secondly, as well, because um there's a very strong distinction between stuff that habeas doesn't do, which is like just completely making stuff up, or you know, people can always use our tools in such a way that they can literally click on a case, they can see where it came from, they can validate that case and its effectiveness. But also some areas where habeas might still fall down or have issues is, you know, um, if it's a particularly area comp a complex area of law, maybe habeas won't be able to find too many precedents on that matter and it will sort of flag back to the user. Or sometimes it might also throw up a bunch of authorities that are relevant, but maybe it's missed one case just due to the nature for the surge from the query that users conducted, that is still very salient and and to that matter, and that which habeas kind of should have revealed, that can still happen. There's chimps in the armor. I think, unlike other, you know, some legal tech companies have been very dishonest in the way that they've marketed some of this stuff and like promised users certain things. So we just try not to do that. And that's why we also have qualifications around because there is still a generative element, which users can turn on or off and they can use the platform as a pure search and just have no generative element as well. But for elements of the platform that include a generative element, there's still a capacity for even small things, like habeas might reason by analogy to construct analysis on something. And it might not even be wrong, but it might not just flag that to the user. So someone being aware of some of those smaller things and nuances of how to utilize Sleekle AI is important. Um so yeah, hopefully that's a bit of a longer-winded question, but I think it leads to these broader things around the complexity of the law itself as a moving vehicle, how we kind of def defend objectivity and subjectivity and all these sorts of things. And then a broader thing though is just to say to an architectural level, you can actually run evals over your basic search stack. So you can ensure that, okay, we're expecting to see at least these three key authorities return somewhere in a search stack. You can have evals that you build for um that, which is also possible. They don't fully, I don't think, solve the problem in its entirety, but they're very useful for building out a base level of like faithfulness and kind of ensuring uh which authorities are retrieved and why and stuff like that. Yeah.

SPEAKER_01 48:55

Yeah, totally. No, that's really helpful. And it does a great job because it grounds a lot of the the results in, as again, as we said, judgments, ostlee case law, you know, the statute itself. So that's great. You've made a distinction that I haven't heard before, which is you've got the fake citation that can come out, but there's also one where other AIs can over-interpret the relevance. Is that second one? Is that a problem that can be harder to solve?

SPEAKER_00 49:21

Yeah. And I also think, again, this relates to on a human level as well. There's there's so much decision making and strategy at play. So the way that like one litigator, even in a totally removed from AI, would apply a particular authority or um a judgment or whatever it might be that they've uh found is quite different to the strategy or approach someone else might might take. And even in a firm setting, I've seen things like with habeas, where one partner um agrees broadly with the output that habeas has generated, and in particular in situations where it's like a strategic recommendation to a client, and another partner has a different approach, and they say, no, we should actually have an even more hardline or you know, a different recommendation here. So even at the human level within firms, you have disagreements, and then um, I think it's more just that awareness of like, for example, when habeas, someone can configure an assistant, they can configure or prompt the model to be like to represent a certain role or to undertake a certain viewpoint um or ontology that they often can't do as easily with traditional AI platforms. And that means that that then informs the analysis habeas is given, giving and why. And they can even kind of like a role-play different scenario. So it's like, you know, um, look at looking to this judge's history and then um, you know, how they tend to rule on matters, what factors are at play, and then um from that sort of pre-game or war game, the scenario in court, according, and you could set up a few different chats basically to analyze that right. So I think people now, most sophisticated practitioners are also moving beyond this idea of, oh, I'm just submitting a very basic query to the chat. And if this AI doesn't give me the perfect barrister level like response that I need, it's failed, and I'm just gonna reject AI in its entirety. People are seeing that it's it's useful in ideation, the prompt still still matters itself, the platform you're matting utilizing still matters and and and all this sort of stuff as well. And then how you work with a tool, I think, like and the the mental framework becomes really imperative. And that alludes to the discussion you you kind of brought up before, which is that some people have utilized tools in the past like ChatGPT, that's happened. I think with one of Leap's products, it's happened, or why um copilot, whatever. And it's not so much product specific because if they were utilizing, I think, those products in an awareness of, you know, okay, what can and can't it do? What sort of functions will excel at? What do I need to check and verify on my own end? They shouldn't necessarily be running into those same problems. So it's again a bit of an awareness of how the tool works and why am I using it and when, and what's my level of you know, verification need or things like that as well.

SPEAKER_01 52:20

Yeah, totally. No, that's that's really helpful. I guess we're in an age now where you know there's so many AIs popping up, and there's heaps of new ones that I'm seeing, and a lot of them just kind of use a lot of LLMs out there. Where do you think the moat is? Abias and some Australian legal tech platforms had a first mover advantage, which was helpful by several years. Where is the moat specifically?

SPEAKER_00 52:42

It's a complex question, yeah. I think more broadly in technology itself, there are still moats in terms of these moats people might talk about at the moment around taste or design fluency. Um, I think it matters a lot more how your front-end experience, your UI, UX decisions matter quite a bit more now. And this is kind of contextualized by the fact that, you know, with vibe coding, but also like just agentic coding, um people can try and clone or kind of rap roughly replicate existing products, you know, more quickly. But the tech stack moat itself is kind of like not in in uh in as much of a strong position as it was. And we know that this is changing in the market broadly. That's not to say that I don't think companies still will have distinct technical modes. They still have to a degree, um, but it does lead to the insight that you know moat might exist in other uh capacities for companies now. So one might be a unique distribution strategy that they have or a marketing advantage that they have that no one else has or that they're building up over time to it to a certain degree, of course, brand name still matters and plays a role, although it's not always as strong with a mode as it may once have been for larger companies in legal tech. I also think the capacity for um you know companies to tell stories, effective stories and narratives, is you know another area. Um, and it's one where like some massive legal AI products have sometimes succeeded. In lieu of having the best products, they've been able to smash through certain markets because they were very good at storytelling and telling a narrative of where they would be and where the product could go and and why senior lawyers and decision makers should should listen to them and think about their product. Um so these are just like a whole bunch of things I'm kind of, you know, looking at and thinking at, but I do uh thinking about, and I think it's where those smaller things now, where people have a lot more optionality in the products they choose and thinking deeply about what would lead someone to stay with yours over another and commit to yours. It can sometimes be those smaller things now around taste, output, you know, sophistication of response on the technical side, which are determined by a whole bunch of sort of micro factors. Um but I really think, yeah, more broadly in in long term, in the long term, we're seeing a lot of motes probably get eroded at the moment and a like sort of reshuffling of that. Another thing you could speak to is maybe identity and things like that and who you appeal to and why, and and having a really clearly defined ICP probably matters even more now, where then you build trust relationships with certain types of communities, and um that helps with brand name and and cementing yourself in a certain position as well. Um, but yeah, what are your thoughts? Because it it's a very tricky time, you know, and with AI, I think it's brought a lot, not just legal, but a lot of markets into a lot of chaos as well. And you've got stock price of some big legal tech companies going down a lot, you've got other companies getting hit like at last year. So yeah, what do you how are you approaching that in your kind of business context?

SPEAKER_01 56:10

Yeah, it's a good question. I think you need to kind of have multiple tools on hand. You don't want to overly rely on one. My take is GBT is a good starting point. It's conservative in its outputs, can kind of help you. A lot of the AI is around framing. So how do you ask the questions? For example, you know, you ask a simple question, that kind of feeds into the answer you'll give it. But sometimes I'll put together a document and say, you know, if you were opposing counsel, how would you fight the arguments or what would be the things that you'd be looking for? So I think that's one aspect of how you use it, having different models more broadly. But again, you've got the user data, right? So for established platforms, I mean legal tech, it's a bit of a networks effect, right? The more data you have, you can use that to either train the model or to understand from a product stack and development point of view how to improve the usability. And yeah, without knowing the technical details of how these platforms work, it's creating the flywheel effect so that you can build things. And if you have integrations as habeas does to the case law databases, I know you've got Jade AI as well. If you can stack in as many of the integrations as possible and then create, as you said, a proprietary narrative, a proprietary story to kind of tie that all together and create the flywheel effect. Um, I think that's that's helpful.

SPEAKER_00 57:28

Yeah, for sure, for sure. And um, you know, at the same time, while of course, um, yeah, I think AI is probably fundamentally changing certain market dynamics, um, it is changing the dynamics around software as a service products in in quite strong ways. It might also have this forcing function as well that um can help with businesses that are small to medium-sized businesses as well, um, which is that a team of 10 or 20 people might be able to make a business that does quite a considerable profit because their marketing, their growth is for each individual employee or contractor who worked within the business is aided by AI, right? So they're quite AI native and they have a sophisticated plan for how to use it, when not to use it, um, to kind of increase individual people's leverage within a business. I think there is an interesting bull case or software businesses that survive in this niche of not just trying to rapidly exp build massive teams and raise massive amounts of capital, but actually build in specific niches, keep niching down on particular product areas and particular types of communities and solve particular types of problems really, really well. And so, yeah, like there's there's some definite positives, I think, in SaaS, despite the fact there's also some big threats to large established software provider companies that are going to have a lot of thinking to do in terms of how they restructure their business.

SPEAKER_01 59:00

Yeah, it's all changing and it's you know, it's moving so fast that you have to stay abreast, you have to stay up to date. So moving on to the broader AI and legal profession, big firms have innovation teams, they have big budgets, they've got IT teams and stacks. You know, you've posted that some small AI firms they manage to compete with big firms despite having little resources. What specifically do you think big firms are getting wrong? Where can they improve?

SPEAKER_00 59:29

Yeah, look, I mean, I guess you know, the framing I'd say is not so much that um big firms are necessarily getting things wrong or that I'm in a place to, you know, dictate around what they can improve. But um I think it's more just different like risk appetites and you know, this this introduces a lot of complexities at enterprise um that sometimes smaller businesses can move a bit more quickly. So yeah, at an enterprise level, you've got security concerns, you've got concerns around who's going to stay around long enough in the market, who's gonna be here in five years' time if we do adopt a new technology. Who do we want to work with in that capacity? Sometimes enterprise buyers would even be okay with you know a product being worst and other offerings on the market if they can reasonably assume, oh, well, we think over time it might catch up. Um the the the framing, and I think um, you know, from an enterprise perspective as well, you've got like lots of different people in the business with different competing priorities and ways of evaluating risk, right? Um I think, yeah, it just introduces some of these complexities. What I think that means where small firms who are leveraging AI successfully have a real advantage, or even like boutique litigation firms, I think, that are still growing quite rapidly. Like there might be 20, 30 people and they're looking to scale up to 40 or 50, um, but they're utilizing AI in a very small way or AI-native platforms. And the advantage these types of firms might have is just that um they're more willing to make a decision, make a call to try out different platforms much quicker to actually experiment and and dig deep into the pros and cons of different technologies and basically test it out on their own user account of self-in, you know, self-serve function like what habeas enables. And then they're also empowered at the at the level of if they've only got a small team around them, each, as I was alluding to before, each person in the organization who knows how to leverage AI as well as a combination of other tools successfully, um, can can can do more work and enhance the quality of the work that they're doing as well. So I think as well, yeah, it's kind of that thing of I just see a lot of on a on a basic level as well. So prax might just also be more hungry. Just be like, you know, some people will say to me about habeas, oh yeah, look, I don't care what this larger enterprise is using, I just care what works for me as an Aussie lawyer that just wants the best quality tool set out there for the work I'm doing. Um, and so yeah, I think a few different factors at play there. But you know, that would kind of be an overall kind of summary. And then broader long-term implications is that probably a lot of firm structures will will change quite rapidly in the next 10 years. You also see some smaller firms that get, I don't know, acquired or um be threatened by AI as well, especially if they've got older principles. And once the principles move on, uh, there's not enough institutional knowledge or there's not enough depth in the ranks. Um so those firms are also going to have experience issues. So it's an interesting fork in the road for like how people choose to run their businesses. And I guess one person we can again with your previous interview with Damon, and how he's been able to build a practice that is deliberately small and tight and like focused on very particular types of issues and disputes and litigation work, right? And then business owners from that can make decisions around how they want to scale or if they want to deliberately remain lean and more profitable as well.

SPEAKER_01 1:03:07

Yeah, absolutely. I think like a tool like habeas is probably a boutique's best friend, especially as they're getting started. Not to say that it would replace lawyers per se, but it just drives efficiency gains and it allows them to scale and grow faster. I guess in terms of you know the billable hour model, that's obviously changing or can change with AI efficiency gains. How do you think that impacts lawyers? And is there, you know, eventually a model that replaces it, or do they just serve more clients and be able to do more work?

SPEAKER_00 1:03:37

Um, yeah, I think it's already impacting lawyers in a fairly sizable way. Like I try and inquire with our clients about like how they're billing and approaching different things and what strategies they might take. And I think it's quite case-specific. But we're already seeing that, you know, we've got different models emerging, particularly in the US, but even in Australia, where people like, you know, you've got companies like Legal Vision but charge on a more of a subscription basis or introduce different models and have been, you know, to my knowledge, successful in doing so. Um, and you've also got models in the US, like companies trying to raise money for AI-first law firms that deliberately embrace certain kinds of automations with humans in the loop and checking it at a certain level, right? So I do think we're already seeing massive shifts, I think or continue to change. I just think at a certain level, it's a it's a unique billing model. It's one that's kind of occurred in pre professional services. And I do think AI broadly at least changes it or um means that people have to be looking at other ways to justify even the value they're creating within a certain hour period as well, or whatever it may be. So yeah, I personally think it's it's a model that's going to get lots of revision over time. And even for firms that, you know, continue to work on certain billable hours, they'll just think about how they structure, what they bill for, um, how like what degree of quality, sophistication, et cetera, with the work they're actually doing in the output and how fine-tuned it is. Like it has a lot of implications. So it's not not necessarily like everyone right now is going to switch or or change their model, but I think a lot of firms will be looking to be first movers in innovating different types of models that work better. And then the other side to this discussion is just thinking about consumers and people who are skill, like routinely underservic, people who might want to pursue legal issues, don't have the money at capability, so they just force the wayside. Access to justice stuff with certain people who, you know, um may not be able to pursue claims, or um, people as well will see, you know, more self represented litigates as well, who will want to utilize AI platforms in one capacity or another, both with some positive effects there in terms of if. Empowerment and being able to understand the law there and some negative potential impacts, like what we've seen, like with the Fair Work Commission, where sometimes people can sort of be misled a little bit if they lack that deep knowledge of what constitutes a successful claim or something. So we often have this tendency to think about law firms and billing, but of course it's a market and it's like, what do everyday people want? What are everyday people in both small business and medium business contexts increasingly going to do? I've already seen a trend where I think a lot of small businesses will utilize Claude or some kind of tool that they can pay a hundred bucks a month for to deepen analysis or to get to a like first draft of a contract and then get it checked over by a lawyer. But again, that in and of itself is a very different model that we're already seeing in terms of how people and smaller businesses empower themselves. I think for now, larger businesses, the reason why they hire a lawyer or a legal firm is quite different anyway. And, you know, they'll continue to nurture relationships, but it might have broader upstream impacts there as well, actually, because what I'm seeing is like in an in-house context, a lot more investment where companies might prefer to invest more in their in-house functions and rely less on, you know, external funds. So that could be another dynamic that we'll see play out. So yeah, lots of implications with billing models and then changing market structures as a result of AI that has kind of leveled certain playing fields and um you know change the ways people are just psychologically approaching legal problems every single day, really, all sorts of contexts.

SPEAKER_01 1:07:39

It's a great point on how they approach things. And I know you're deep, you're passionate about the ethical side of things. So I guess that's a good segue. Do you think legal AI more broadly, does it make access to justice better or worse? Does it help people or does it just make lawyers faster?

SPEAKER_00 1:07:55

Um, I think that it's it's a double-edged shore there, but I think basically proper properly architected legal AI definitely in the longer term probably represents a net benefit for access to justice. Right. Um, but it's more about us as a society in Australia in particular, building out the infrastructures and frameworks that support proper rollouts or proper funding of, you know, for example, community legal centers or people who provide services on larger scales. Um and I can give you examples of that just in a context of, you know, first of all, a CLC might even take a look at something like habeas, and then uh they might be very early on their AI journey and sort of, oh, what we, what, what options are out there, why would we pit punting over another? But even then to get the like funding to access a solution that could then rapidly increase the amount of people that could be served, the amount of matters that could be processed and triage, and it doesn't matter what tool it is that we're talking about, yeah, they'll need the right level of external funding, whether that's through government or whether that's through work that they're doing or some kind like, and that's where the trickiness arises for a lot of organizations. And the second thing, you know, in Australia that we'll often do is I think we sometimes are again quite late to, you know, adopting newer technologies at at speed and and scale. And there's a massive opportunity here in access to justice with using tools like habeas, absolutely. Um, but I've actually even seen some pushback from CLCs in the past who maybe just don't get it or kind of like, oh, well, you know, why AI? And could this also decrease the human element of what we're doing? Which I don't think it has to. I think if it's sophistically leveraged by someone as a tool in their toolbox, um, and you've got real people who are also making calls in terms of ethical calls and how to represent advice to a client, we can absolutely mitigate that risk at the same time. Um yeah, the the second thing also being just like I'm always interested in leveling of playing fields and and looking at, you know, to what extent is someone able to access and knowledge, resources, relevant information. Um, a self-represented litigate should absolutely, you know, ideal world, have access to the same research tools and stack of a large firm. There's no real strong argument against that. Um, and so it's kind of more broadly like we've also had a you know a situation where a lot of people have just like been priced out of certain softwares in the past. Um, and so um, yeah, again, if we can change that, I think AI already has changed that, as I was alluding to before. That's a positive thing. We also need to be, of course, slay, you know, aware and careful of what's going on and the fact that there's this kind of whether when a new technology gets rolled out, there's a lot of teething effects. This includes, for example, self-reps who've you know submitted really massive documents or whatever, and judges have gotten annoyed, or um, there's been all sorts of you know issues talked about in court contexts. This has impacted no self-reps and noise to some degree. And yeah, it's not being unaware of those things, but it's talking about it in a in a grounded way of well, okay, well, what can this technology do two, three years from now as well as even what can it do right now with proper and effective rollout? And the government should be absolutely looking at these things and prioritizing them. I know that the Labour government's done some great stuff in terms of AI initiatives, um, but not just kind of taking the default path of like, oh, well, let's roll out something from Microsoft internally. Let's look at other solutions of like how can we fund organizations that are making a difference with AI? How can we potentially help with better funding community legal centers so that they can adopt AI, all sorts of structural changes that need to be made to really have an effective societal level shift, which is what AI should be able to provide in theory, right? Because we're talking about a technology that still for its faults in legal context can still be extremely powerful, right? And I've even seen settings like as a final note, what I'm seeing is I've seen a few settings where self-reps, sometimes who have a legal degree, actually, or sometimes don't, but I've seen them, yeah, utilize tools in court settings quite successfully as well. So that's often not talked about. People just look at the negative stuff for like when a lawyer fails or a self-rep's done something that's not by the book. People are looking less at like the successful case outcomes, especially in small claims courts and things like that. And it is happening a lot more. Even self-reps successfully like representing on matters where there's like a decently large legal team on the other side. So that also fascinates me and interests me deeply. And I I've only been able to come up with a few theses of why like this is happening. But sometimes it's also that some lawyers, even professional lawyers, are overstaffed or that they've got too many matters going on. So if a you know a really competent and thoughtful person who sometimes, again, might be self-represented or they might be working with legal aid in some capacity, can also then use AI. Um and a few self-reps I've seen have who have figured out about habeas, even though we've never really marketed to the consumer market and have done quite well. So um, yeah, again, really interesting. And uh again, this is like kind of a side note, but just something that popped in my mind is like, you know, a lot of legal matters are quite cut and dry. So like I, you know, this guy was not a long-term customer, but there was a guy who I think um it's not our target market, but he used habeas to resolve like a traffic, you know, error and basically just contested that look, like I wasn't using my phone um in this setting. It was clearly a mistake of whatever technology was being used to sort of capture him as he was driving. And that's just a simple example. They're not really our long-term market, it's not necessarily the best type of from a business perspective, but I it actually um small matters as well. There's longer extended matters which are going through courts. Yeah. So lots of ideas there. But basically, bottom line is yeah, AI can be an incredible vehicle. It's also like how we thoughtfully deploy rollout and which initiatives we back, and then even at a government level. Um, and I think for us, for our part in habeas, we're glad to also be helping people of all backgrounds and and legal backgrounds as well, who who better quality legal research and better quality knowledge so that they can empower themselves in a in any setting, right?

unknown 1:14:37

Yeah.

SPEAKER_01 1:14:38

It's a great point on the justice gap. I know you've mentioned 80% of legal manners in Australia don't truly get the time they deserve. People either resolve it themselves or they just go unrepresented. I mean, you see stats out of the Fairwork Commission and Fairwork claims are going through the roof on the other side of things. Uh, I think there was a court in Victoria I heard on one of your other podcasts, they've apparently blocked ChatGPT in generative AI. That can cause, I can see that the side of the judges where they don't want people to plead all this AI splot. They want actual properly pleaded arguments. How can say, like, someone in that position, can they use habeas? And would you consider, I know it's not your primary target market, but given you know how effective it is for self-represented litigants, would that be something that you would look to kind of promote or not really?

SPEAKER_00 1:15:24

Um, I think it's just it, yeah, it's being careful around our preferred, like, first of all, when we like it's always been my endeavor, you know, with with some challenges along the way to work with people in different legal contexts, whether that's firms, but it could also be community legal centres, because you know, people who are more specialized in a particular area and practice area know more than me. And then they'll kind of know, okay, okay, once we've got something like this at our disposal, like what is the right way to leverage it all that out? I've also just like kind of advocated for this. Yeah, I think at a at a base level, the combination of a legal officer or someone using habeas in conjunction with their own knowledge, um, that's been a focus area. But I also think, yeah, inevitably, that's just an age where once a product is out there, people, every everyone's as long as they're actually properly paying for for a subscription, etc., has got the right to utilize technology as well. And so, yeah, I think it's just um it's something that you know we're keeping an eye on in the market. But at the same time, we also want people to be strongly aware that, you know, isn't a replacement for a lawyer and those sorts of things and the distinctions between what is actually happening when you're um going getting advice from a lawyer versus what's happening if you're using an AI platform in any capacity, whether that's a legal specific platform or not. And just so that people are also aware on that side of like they're conceptualizing habeas, it's just like, oh, this is just my lawyer now, like it's not quite right, right? So yeah, I think broadly we've also just tried to work with various different facets of the perfection. And then over time, that will probably have a lot more network effects, which is that some people, yeah, we'll be able to utilize it in a sell-sort capacity, but ideally we'll probably be able to make the most change at a structural level if we can get through to more enterprise, but then also community legal centers, right? Um and this brings me back to the point of like um yeah, the sometimes the fears that people have, which then may leads to delayment in adopting newer technology or or thinking about it. And then um, if you're a CLC and you're kind of like not using more modern tools and thinking about your tech stack, you're probably um doing a disservice to certain clients in a in a more abstract sense because there's a limited matter amount of matters you can always actually be taking on. I mean, and appre if efficiency is not just a buzzword, it can have real implications for like how people are being helped and um yeah, when it's done right again. So yeah, that's a a bit of a summary there. But of course, we've always been focused like from a business case, probably a bit more focused on the B2B, but because Havias enables that self-circacity, people can come along, of course, and they're not restricted from utilizing them at all. And I like that general idea of, you know, um trying not to impose as many restrictions on technology in general, um, because I just don't think it's makes sense in the context of what's happening in our current era and like how people are empowering themselves, and then also how professional services workers are doing lots of cool new things with AI. So yeah.

SPEAKER_01 1:18:44

Yeah, no, that's awesome. I guess this last question on this section, there's been a huge proliferation of like AI used for self-represented litigants, and you know, courts are not overly happy with that in some contexts. Does AI change in the sense, let's say for habeas, where someone who who wants to use it is able to to get a more clean output from it or to guide them? How also do you think the courts, how do they view AI as we look in say the next 12 to 18 months? Do they go harder on it? Do they just ignore it? Or are there ways to kind of clean the output so that people can use AI and it spits out something that, you know, is not just pure like as they call AI slot? Sure.

SPEAKER_00 1:19:26

Well, I mean, there's a few, yeah, like that's a multifaceted question in terms of how you'd go about solving that problem. I think at the first level, court guidance just being like moderated, where, um, and kind of aware of the fact that, yeah, the technology exists and it's out there. And and sort of, I don't think we'll necessarily see any kind of like attempts at blanket bans or something. But publishing really rigorous guidance, and that includes like that guidance, having a strong awareness. Like if you're publishing guidance in 2026 is just like, oh, AI can always hallucinate and you know you can't trust it, that's out of date. And it's kind of like courts need to be following that. And that, you know, what one thing I'm you know saw recently, even there's like um some core websites in Victoria that were looking at um resources for AI research tools, and habeas was mentioned up the front there, which I really appreciated because you know, people click through those resources and check out different tool sets. Um, but you know, so it's also about then, you know, on a second level beyond the guidance and the regulations that have been put in place. Um ensuring that, yeah, people are directed, made strongly aware that there are technologies available which can do a better job than just like maybe using a default platform or that are custom built for that purpose. Um, I think more broadly as well, reviewing ways that we can uh improve efficiency of court processes should be a whether that's related to AI or not, just is something that Australia needs to do. Um, because like, you know, it's kind of like we don't want to operate in this fantasy of pretending that like pre-AI everything was going fine in courts. Like it just certainly wasn't. And there's lots of people who deal with matters that go on for way too long or um, you know, due to admin things going on. And then yeah, I think I I don't have a sort of perfect answer to this question though, because I think anyone who's in a a court context and a judiciary context, it's like quite a difficult, you know, set of factors to be grappling with, both the positives and the potential things and risks to mitigate. Um and I know it's it's going to be a tricky job, but I really hope that Australia as well, as a nation, we can take a more forward-looking approach here and then think about like what are some broader ways we can bring in, whether it's AI or other forms of technology, to do a better job in courts themselves. Like one example recently, which is like in the news, is dictation and like a recent like dictation software that was being used by a lot of um in a lot of court settings. And the company, I think, folded or you know, had some kind of structural issue. And we should then be looking and thinking of, okay, well, we've got some awesome AI dictation services now out there from you know, technical providers who specialize in that niche. How can we roll that out in different court settings so that we can capture context aware, voice-aware stuff? And then we can also then have humans to check output or to, you know, have additional safety guarantees and accuracy guarantees. Like we should, we should really be thinking about how we can even cost cut cost efficiencies because sometimes the amount invested in those softwares that is that are that are 10, 20 years old is quite substantial. Um and yeah, just basically taking that less of a fear-based approach, I think, whilst also really being strongly responsible around like um how how people even educated the base layer of AI, which is a final point that I'll mat that I'll mention because we've even done that in the context of habeas, which is really give people a strong idea of like what what are LLMs, and then okay, what distinguishes a certain type of product from another, and why does that matter in terms of how you use it all? And that sort of public education, even beyond just lawyers, I think is also extremely valuable. Um, we don't want to get caught as a nation where we're just like several years behind what whatever state of the art is. Yeah. Fantastic and great point.

SPEAKER_01 1:23:26

How we use AI. It's one thing to know about it, but how to actually use it to get the best outputs and to leverage it. So that's awesome. Just want to move on to partnerships and distribution briefly. You know, you've had a number of partnerships, you know, you display smoke ball on your website, and you've also had, I guess, some dealings with Swimburn University, CPD partnerships, and Alpma. I guess how has habeas, you know, approach partnerships and distribution as you kind of lead into more of a like a product-led growth or a sales function?

SPEAKER_00 1:23:54

Yeah, it's a good question, actually, because it's it's always kind of on your mind in some capacity. And I think especially as you grow time and you're trying to uh build a business as well and and and transition from being just a builder-oriented person, which maybe I was for a period of time and just loving building and creating things, but really the way I look at kind of distribution and growth is it's kind of like from a positive angle, if you've got a good product, then you know, you know you've got at least a decent network of people who've really given you gotten to that stage. Oh, I love this or I'm really using this deeply. It's usually indicative of the fact that you could create better impacts on the world if you just understand how to, you know, market or grow something better. And of course, it's like probably the the evergreen skill in growing a business, which is like the sales and marketing functions, right? And that's just there's so many different sub-skill sets. So it's been my goal to try and get better, a lot of those underlying skill sets, as well as increasingly get people on board who can help a bit in various capacities with growth, right? Uh, I think to chart our different strategy and stuff over time, yeah, with the partnerships element, that was something I think we focused on in 2025 and early 2025 initially. And I think it's kind of that thing of, again, just trying to establish some branding and awareness. And it's it's much trickier than people might even realize, but to just get out there to all sorts of different people, create awareness. I think there's this concept of digital gravity, which is like you need to be in someone's mind, you know, multiple times a week, or them seeing your page or your content or hearing about it from someone multiple times, like eight to ten times before they'll even really seriously engage with you. Um, so it's not just getting awareness or someone having heard of it. It's kind of like that compound effect. And yeah, partnerships were were useful in that capacity of getting some runs on the board, getting some awareness, getting some people who already use an existing technology like Smoke Ball to then come across and be curious about your platform. They can help with distribution for people in in different networks as well. So even in a more recent context, we've like, we partner with some sole practitioner networks, like for example, Jason Loft and his community. And they um we introduce like a reduced subscription, but for people who already pay for his community and um sort of firm-building service that they offer to primarily sole practitioners. And um, yeah, it's more about trying to find partnerships where there is like a strong interception with um the work you're already doing or it really fits with your ICP. The cop the converse side of partnerships is sometimes, and that was just a learning from last year, is like um I think with junior entrepreneurs, sometimes you go and partner with too many people, or maybe, for example, there are certain partnerships where someone promises you something quite tangible and then doesn't deliver or doesn't necessarily take it to that next step. And I think partnerships, of course, as well as like a is a quid pro quote. So it's kind of like the more that you can actually offer something uh to someone as a business and the value that they're giving, and it might not just be product value, but it it might also be they've got to make hard decisions around sales volume and who they work with. And, you know, that's where the other point comes in, which is I've been try to just focus on learning skills around how can you build your own distribution network and be smart around that. Again, not an easy thing, but businesses that build a really deep sort of sense of like they have a clear lead acquisition, you know, structure, they have a clear methodology for what they're doing with sales. They have a clear methodology for outreach. Even just basic things like that means that you're kind of building your own growth engine. And then oftentimes particular partnerships are probably more likely to bear fruit or come your way once you've built that out. And a good example, of course, is like the business you work for, which is Law Path, right? And Dom's done like probably stone mining. Work on like you know, I even know Larpath as someone who's used some of the legal services in the past for setting on the company, you know, like how you actually sold in or like the scale. And Larpath hires multiple people in sales and growth functions whose job it is to build in these repeated systems, right? And that's kind of part of the growth distribution engine that has been built internally and other external partnerships have come along, you know, investment from banks and things like that as well, which is incredible, but probably more as a result of that initial growth engine accumulating, right? So yeah, I mean massive learnings around this stuff, and and nothing of it is super easy, but it just probably reflects that this year we're being, you know, with any particular partnerships would do a quiet choice. And then the structure I'll probably think about in depth. And, you know, there's one business in particular, but there's a couple of businesses that we'd love to partner with in the future just from a data perspective or um expanding the access of the different things habeas could search, for example. But yeah, it's not as much for priority focus for 2026, maybe. Yeah.

SPEAKER_01 1:29:09

And I guess, you know, at a very high level for habeas, you know, obviously you've mentioned, you know, getting out into the real world conferences, events or something that you hadn't done and regretted at a high level. How are you looking at the sales function for habeas? So you, you know, obviously you've got the the pilots that you're running, which are cool. Is there any kind of you know, major strategies you're looking to kind of move those pilots along and reach a new distribution of customers, or is it just kind of promoting the brand and word-of-mouth awareness?

SPEAKER_00 1:29:37

Yeah. I mean, without going into too much debt, we've just probably got a more consolidated strategy. And again, it's just contingent on a few things like capital to a degree and what we can put in as well into marketing as you bootstrap a business. But yeah, we have a fairly well consolidated now, at least idea of like we're producing a lot of regular content which brings eyes to the business in different ways. We're having chats and conversations with people, you know, like yourself across Australia, who, you know, kind of create different kinds of awareness and podcast settings, etc. We're we're running, you know, paid ads and things like that, as well as different parts of our strategy. Um, and yeah, look, more broadly though, it's just there's been a bit more of a proliferation of network effects that I've noticed this year compared to last. And that's been helpful because there's all sorts of different ways to kind of market and think about go to market, and it's really complex and it depends on a lot of external factors. But having endorsements from people in the network and encouraging people even to share testimonials if they haven't already. Um, because I oftentimes find that I'll ring someone up who is a customer, but I just haven't personally built a relationship with that customer. They use a self-serve and learn so much about their use case in a particular industry. And then that qualifies my ability to talk to someone else in their practice area about the value that they could get out of the platform. And it also means that all people are likely to kind of go share their experience in different settings. And where that really hits home is I'm seeing people who are in literal enterprise context, like subscribing to habeas in a sell-so capacity and then bringing it back to their ops or knowledge leaders, being like, I had a good experience with this platform, which again, for you know, sometimes smaller businesses where it might be hard to get a look in or where that business might just have heard of the hyped up products in the market and that's where they've gone. That sort of thing as well is very powerful. Um, you know, we've even got referrals from Solpracs using the product to larger litigate litigation firms, which then led to pilots or arrangements occurring, right? Um, and so yeah, it's a it's a more longer-term growth thing that is just born out of sheer, probably to a degree, force of will, but also network effects now proliferating. Even Damon Murdoch, who you mentioned, you know, referred several barristers who are now using the product. Um so we really want to be building on those things because we don't see it as necessarily like if we just try and copy a strategy of much like extremely well-funded businesses that are really looking to exclusively target enterprise or whatever. There's not much validity there. You're just kind of playing a game that other people have already played and are um going to have more capital to invest in, right? So, yeah, probably a lot of things that I that I can't mention as well. But um, this year, that's where like there's probably a more sophisticated approach overall. And we're building more systems in the business that can help us scale out. Um, but you know, don't want to pretend that at the end of the day, this is really the the magic and to a degree science of why you know business itself is so difficult and why most businesses in startup contexts do fail, particularly in BC contexts as well, um over the over the long run, like for especially for newer SaaS businesses rather than like services businesses. So, yeah, it's part of the excitement as well, is figuring those things out, finding your unique um points of difference. And um in in habeas' case, I think it's cool that we've got a bit of a a different narrative that sets us apart um from a lot of other providers or providers that have sort of come out of the wings later on once the need in the market was validated and stuff, yeah. Totally.

SPEAKER_01 1:33:34

That's fascinating. In the US, you know, why Combinator is now backing AI first law firms in the US. Do you think that's something that's coming to Australia? And does that represent a threat or an opportunity?

SPEAKER_00 1:33:45

Um look, I I actually don't think so. For me, I don't kind of see it as either or. I just see it as like the market is going to move, like especially from uh well in well-capitalized countries like the US, where there's just like a real flow of capital into anything AI right now. Um, it's a less risk-averse market. Sometimes just um larger investors will, you know, go into even sometimes more unvalidated spaces if they think there's real thesis or reason to be in it. Um and yeah, look, no, I think that the same thing has already emerged to a degree in Australia anyway. So even firms that don't necessarily call themselves AI-first law firms, but like even in the legal work, for example, your business law path does, right? Like so much of it is informed and by this complex interplay of not only the lawyers who bring their own expertise, but also certain AI and automation functions already in the business, right? So um it's it already exists to a degree, and it's just kind of to what extent the marketing or the model itself is a little bit different. And in the case of some of those businesses, yeah, they're coming out of the US, I think you'll probably see echoes of it in Australia for sure, or you might see a few law firms that are gonna exclusively target the idea of no, we're completely AI native or something like that. Um but I just don't think it like it will occur in the same degree, at least for the next little while. Then in terms of the broader implications for SAST, it's kind of like it's really just the type of business you want to be, and they each have different scaling laws and things like that, right? It doesn't mean that technology that assists, because I think ultimately all of these firms will either build their own internal stack or work with an external provider anyway, um, likely wear providers of some kind, or they might even adopt a plethora of tools. So it doesn't necessarily mean it's a bad thing. Like, for example, we've assisted a lot of firms that I would call quite AI first anyway. And so in their in their structure, their output, they're leveraging habeas a lot. And so it's kind of like these definitional things as well. It's like where do you draw the line in all of this anyway? Is is is different, I think. Um, and yeah, like I do see it as like there'll be some pretty big transformations in terms of how how certain work is done as a result of that. But do I see it as like a long-term existential threat for just like any and all SAS tools, including habeas? Probably not as much because it actually creates some additional buy-in and demand. And secondly, there are a lot of firms, particularly in the Australian context, that will have no desire to move towards an entirely different model. They'd like a more progressive, gradual approach where they can adopt software and they can also continue to employ the people they love working with and build out their team. And um, they're not going to be approaching it from the same narrative or thesis. So yeah. And even yeah, we should also qualify this even further by kind of saying that, yeah, you see this big stuff in the news, but then you actually go to the day-to-day realities of like certain legal firms, and there are still a lot of legal firms here in England, in uh in the US, where it's like still the technology of these firms is still much further behind. So they, you know, still some firms that may not have even graduated from like written documents and don't even have a some kind of basic DMS system. So still so much opportunity across the scale for modernization and um, yeah, all that stuff, which sometimes the bigger news stories kind of obscure that that's still the reality of quite a lot, quite a lot of legal work and firms.

SPEAKER_01 1:37:24

There's, I guess, a lot of space and opportunity to kind of, as you said, kind of combine, automate, and digitize, you know, pre-student management, softwares like Leap around practice management and billing, trust account billings, and then obviously where you come in as well. And it's interesting to see as well, obviously, last few months, you know, some of these AI companies, it's been a bit of a correction, even locally. We've had Wise Tech, Atlassian, and a few of these companies retreating. But yeah, my take on it is as long as it has a good use case, has a product market fit, it's usable, it's sticky, it's part of the operations. So something like habeas, regardless of how valuations and public markets play out, it's a strong product and it has a real use case, and that's something that you can use to grow it. Now, I guess you've bootstrapped habeas to this point. Is that a pure philosophical choice or is that just circumstance?

SPEAKER_00 1:38:13

No, it we it certainly isn't circumstance in the in the sense that you know money in the AI space is available if you want to go kind of get it and go down a VC funded pathway. For me, it was more like um, I think, first of all, from uh when I started the business to now, many things change, right? Um in your own personal learning, your own confidence in what it is you're building, um, your own like business skill set, and I'm sure will continue to change, right? Like in three years, you'll have learned a whole bunch of things that you'll feel like, oh, if I just knew that, you know, that time. And that's just the way time kind of works. Um, and one of the things there was more that firstly, I was very probably enamored when I started with just this idea of like being quite self-built and home growth. And I probably didn't fully anticipate the depth of the sales cycle, particularly with enterprise firms. Like I knew it would be significant. I didn't know at that time how significant it was. And so I probably should have built a smaller firm network and market earlier. But again, I don't necessarily regret that because it did a lot of stuff for improving the product to such a good point where a lot of people then picked up on it when it was released for self-serve. Um, and how that links to, you know, capital and and those ideas of fundraising was again just kind of like I was trying to build it out initially as more of a lean, bootstrapped business by design, trying to improve my own skill set awareness of who it is we're even selling to, why, like what does a product do? What's our points of differentiation, all these sorts of things? And, you know, is kind of uh probably something we also didn't want to delegate too much control to in a VC context, at that point in time anyway, um, knowing that there's certain timelines imposed upon a business, there's certain expectations. Uh, the VC model just come with like a lot of inbuilt assumptions about like what a business should be and where it should get to. And I think now I've got a deep network of people who like I've had several lawyers who've um uh like offered to invest or wanted to invest in habeas. And so I think now we're looking at from a different angle, which is like if we were to raise just who can we do that's like at a unique point of difference in the market, how can we make it something that fits with our brand and images? Maybe it's building a syndicate of people who um really already love the product or just love the idea of like Australian first legal tech and like retaining some of the IP in this country. And there's a lot of people who who can relate to those narratives or who already use the product and are you know successful um in their own right in business context and have spare angel capital. So yeah, there's lots of ideas there. And also, um, yeah, definitely that means that this year a few things might change on that front. But we've also raised a little bit of um angel capital already in the past, just like very minimal, but that's helped with pretty much like, you know, not with any massive equity ramifications, but yeah, that's helped a bit. Um and then in the future, if we go down that path, because I think now as well, it just kind of I'd have a very strong idea of where to put that money towards. And most of it would probably be on the growth function since product is something we've struggled with less and product market fit. And yeah, like a few important decisions, which I'm probably weighing up there at the moment. But I don't think it's necessarily a bad thing as well, that some young founders, when they're like 22, they go and raise quite a bit of money and then they get burnt and their product fails, or they just really didn't have the accumulated experience. VCs have taken upon, and that makes them quite toxic as well to a VC audience, or particularly if that founder like flouts to the amount of capital that they raised and builds their brand out of that without having built a successful business that can be quite off-boarding to future investors. Um, so I was always just a little bit cautious of that. Um, but I also think I have some learnings, which is like, you know, I probably could have gone and, you know, raised a little bit earlier and stuff, and probably would have been fine. It probably would have accelerated certain learnings and growth patterns and uh would have helped get towards certain goals maybe quicker. So yeah, I think there's lots of factors that play into this. But basically, I don't regret the fact that yeah, you you accrue a lot of experience in different skill sets by bootstrapping a business, which means that yeah, you might um yeah, be better qualified to do a lot of things in the long run uh rather than necessarily raising too early and yeah, not having strong visibility of a prime market fit or whatever type of thing as well. Um, so yeah, that's uh hopefully a bit of a an answer there. And yeah, I also think just even more broadly, people are just in Australia, we just need to be also encouraging all sorts of models of business. So people who intentionally build a you know, business small, that's not necessarily us anymore, but people who are intentionally just looking to build a 30 to 50k a month business with them and a couple of employees who, or you know, even if just deliberately laying those valid businesses, we shouldn't only suggest this idea that you're only really worth anything, or you're a serious business if you go to an Atlassian or a Canva and you've had this massive VC story, uh, because it also obscures the narrative of like a lot of companies that raise VC doesn't go quite right for a number of reasons, the majority, and it turns into either um Capri's returned or turns into a zombie company or all sorts of outcomes which happen frequently, and founders don't necessarily get exits on their return time and the way that they model the business. Um and yeah, so hopefully that's kind of a bit more of an in-depth answer, and there's probably lots more ideas there to share, but maybe for another day.

SPEAKER_01 1:44:10

Yeah, totally. Yeah, totally. There's always a bit of a tension, of course. You know, the prevailing narrative in startup culture is you know, raising capital is the dream. If you know you can scale fast, you've got VC backing and quote unquote dominate the market. But again, I I think yeah, the market paints it as the dream. Obviously, there's a lot of drawbacks from that. You're working on someone else's timeline, your equity gets diluted down significantly, and you know, you're it's a bit of a flywheel effect in the sense that you're constantly dependent on raising further rounds, which you know can really dilute you out. Um, but I think for what you're building and you know your broader philosophy, I like it. And you know, I kind of infer that being bootstrapped actually produces better products, but especially at this early stage, it gives you more control over how you grow the product, how you shape the product, and it ultimately gives you more optionality.

SPEAKER_00 1:45:02

And and even just like to that point, you know, you probably um you can you can trial draw an error in a business and you can look at different models. So just because you're worth one thing doesn't mean you can't pivot to going down a different path, choosing a different strategy if you if you are like, oh, we did the wrong thing and that's fine. And secondly, there's just kind of this I kind of see it as an apprenticeship thing as anyway. It's kind of like it takes quite a lot of like it takes that 10,000 in hours, which probably done much more than that by this point. But in terms of like, I don't mind the fact that um it might go against like if you're just purely looking to build a bit massive business, traditional advice would just also be like just raise a bunch of people who so uh hire a bunch of people who are better than you, raise a lot of capital, who perform different functions. But I actually wanted to go through that learning of having a deeper knowledge of like, well, what is a stack I'm building? Why, like what's a product for this philosophy? Can I build at least some of these different skill sets in these different areas again, which will then inform hopefully long-term, you can make better decisions around who you actually hire, work with, which I definitely made some mistakes in the past in terms of just like people I've selected to work with, um, which which can only come through that that kind of learning. So yeah, lots of lots of pros and cons. And you can also apply that to a future business where you might be like, oh, I'm much more confident now that if I take this model, I understand again how to deploy that capital and how to raise and how to do these different things as well. Um so yeah, yeah, hopefully that that all is interesting.

SPEAKER_01 1:46:35

And there's nothing wrong with raising capital. I mean, there's inherent benefits for scale and advantage, I guess the other side of the business or other side of you know raising is that especially for businesses, you know, who are you who can use and outsource or closely work with, you know, AI like Claud Code or something, that dra drastically reduces the capital cost and the the deployment cost so you're able to get things to market quicker, leverage a lot of free tools and and things like that without being too capital intensive.

SPEAKER_00 1:47:03

Yeah, a hundred percent. That's also where yeah, modern technology plays a big role in firstly when you're yeah, building a new business, not just thinking about, okay, well, how can we gain more capital? But again, there's so many disaster stories. And yeah, I've worked with a business coach been in the past who's worked with hundreds of that as startups and has seen that story repeat time and time again. Or a business just exclusively focused on capital raising to the detriment of some other things and not thinking about, okay, well, yeah, whether it's through automation, whether it's through hiring, whether it's through different types of systems and strategies, how can we get more juice and effectiveness out of the thing so that there's a core system that's built up really well and effectively to achieve a certain function? Um, and that might even extend to like, you know, as you said, in the AI context, it could be like three devs who are very competent in a particular area and who work and synergize together really, really well, might be better than a team of 10 where communication's a mix, there's blockages, you know, and and all sorts of structural issues in an in an orc, right? Um so yeah, really interesting time there. And obviously we're seeing these narratives of like one man billion-dollar companies, which often, you know, I'm not to the extent I've seen that, there's often numerous issues under the hood with those types of businesses that are being publicized. Um, and in terms of their ethics and the way they've utilized Heha, even like what we've seen with this guy recently, who like start up a medical GLH supplement type company and then use fake doc doctors to advertise it. So that's probably one of the more negative impacts of AI there. But the general principle holds that there is probably a lot lot of ways to do a lot more with a lot less sometimes, and just be and that and that can help entrepreneurs who are, again, in all contexts in the world and who don't necessarily live in Australia or the US. They live in you know markets and places where they need to work by themselves for a bit or they need to work with a small team for a bit, they physically don't have any more capacity, and that's again a beautiful thing that you know emerging tool sets can help with.

SPEAKER_01 1:49:13

Yeah, no, absolutely. Well, I guess to kind of close out the interview, I know you like to work in time frames of six to twelve months because things are changing so fast at the moment, but where is habeas in the next six to twelve months and what's coming up for the product?

SPEAKER_00 1:49:26

I think in terms of product, it's mostly refinement. So it's like rather than us just shipping a massive amount of new features, people have like set it, settled into a flow. It's more about trying to ensure that like firstly we have a better onboarding structure, we have a better customer support function so that people who are less familiar with different tool sets or how they can even get the most out of what is already there on the platform. Because I've still frequently talk to people who just, for whatever reason, not discovered that that tool set or dive deep into it or thought about oh, how can I use That in my work. They're just using one function of the platform exclusively, like made of the search engine, and they haven't looked at the AI agent functions, which can be quite powerful. So that's just one element, which is building out our actual customer support, our onboarding in an enterprise setting and ensuring we have like really strong user guides and then documentation and things like that and building that out. Another thing, of course, just making a lot of small revisions to the search quality, you know, the user in the face, things that will just these touch-ups, which we want to, as well as like some new features over time, which, you know, we'll sort of announce as as those get shipped. But I really think I'm trying to go with a counter-strategy to just shipping madly and a whole bunch of codes. Because I think a lot of legal tech products now, they'll have like 20 things on the sidebar. It's like workflows, this, that, that, and it's like 10 separate things. That to me, based on past user feedback, it's quite overwhelming. We've kind of like got a call for going in the platform at the moment. And anything we extend beyond that has to build, be built in very deliberately. And then, you know, in terms of where, you know, with the business and where it's going, I think it's going to be a pretty significant growth period just based on like suddenly things in the business. So many, um, you know, I'm amazed with people who've come onto platform who are in some high-ranking um, you know, situations and legal practices and really deeply experienced people who are getting value, um, which is awesome. And then building up your network effects, building up our unique niche and and space in the profession. So a lot of that will also involve scaling out some more like firm-wired rollouts for certain for certain types of litigation businesses and similar. It will also involve collaborations with more collaborations with universities. So, for example, we've kind of got a grant going with Union New South Wales, but um for members of their faculty. And, you know, we'll need to scale that up as as things go in the future. And yeah, creating more awareness to people is is still a central goal in the next little period of time and like scaling up our outreach initiatives and all sorts of stuff. So yeah, I think in six to twelve months' time, the business like not only, you know, from a pure standpoint of revenue will be uh in a better, you know, our goal is in a much better point in terms of like basically just expanding what we've already got and getting out there to more people and getting more legitimacy in an enterprise context. But I also think um, yeah, like that I see this as just a a really unique period for us where yeah, a lot has happened in the market. A lot of people are just like changing their tone on AI or reviewing things more deeply. So they're not just going only to the big brands, they're actually looking more deeply and say, oh, what else is out there? What can we do? I see that as again as like a massive, like just I I'm always aware of these like attention sort shifts that I see and it that weren't true a year ago. And I think that that again probably benefits us quite a bit um in terms of more people coming, revealing the I mean, even someone like yourself, right? Just going and testing it out, self-help capacity, telling your mates, oh, this is great. Like if you're a litigator, give it a go. Um, yeah, all of that compounds. So yeah, look, I think it's gonna be a pretty exciting six to twelve months. And then with that comes probably a lot more sort of new hires and stuff like that in the business as well, which is helpful, which we've already sort of started with, just hiring people part-time primarily, but then also some more full-time hires for like people like founding engineers and stuff like that that's gonna be important. So I basically as well as like taking certain elements of load off you as a solo founder so that you know the key man risk is mitigated and business is diversified and more time can be allocated towards different functions as well. Yeah.

SPEAKER_01 1:53:55

Yeah, no, absolutely. It's really exciting to see where it's all going and look forward to seeing it grow. I guess just last question are you building for Australia predominantly, or do you see this as a potential global product and other markets?

SPEAKER_00 1:54:08

Um, I see that again, a bit bit contention on whether we a few things of like we still want to um focus more on the Australian market in terms of go-to-market at the moment. But yeah, I see there's like probably enough modularity in the platform that for other common law jurisdictions that probably have traditionally been underserved. And I mean, just a few that come to mind are like Canada and like Singapore and countries that share like similar, not necessarily the same, but if there's countries that share similar approaches to law and approaches to material that's actually out there that's open source or that's accessible as well. Yeah, I think there's enough modularity with a platform, it could be shifted. It's just you have to be quite intentional there. And really, based on my past experience, knowing that, yeah, it's difficult enough as well to sell into a one market. So if you want to expand and, you know, ensuring that there's a like really clear narrative for why you're doing that, who you're doing it with, who you're partnering with. And there's companies like Legal Ready that we can take inspiration from where you know they built out a core market in Australia, but over time then we're able to expand into the UK through partnering with a number of sort of existing legal networks in that space. Um and that company's sort of CEO is Tony Kinnear, and he's had a lot of experience with past business building. So it's definitely something we can look at as a successful model. Um and yeah, I think it's kind of like making sure as well, though, that there's lots of little nuances where country a company's sorry, a country's particular jurisdiction, the way that they approach the law, the nuances, the niche things that work for our lawyers that wouldn't work for them, that would be just not counterintuitive, right? Those are the things you from a product perspective really want to make sure fine-tuned enough that that would work. And that's where we'd probably focus on countries that are similar enough that we can still um adapt the product.

SPEAKER_01 1:56:07

Yeah, or something. That's really cool. And it totally makes sense to kind of master and dominate one market first, and things are changing so rapidly, so that would take a lot of your time. But yeah, just lastly, for anyone listening, whether they be you know a young person or they're a senior partner at a top firm, um, how can someone get in touch with yourself and habeas to kind of give it a go?

SPEAKER_00 1:56:27

Yeah. So obviously, I'm firstly contactable via LinkedIn at sort of Will McCartney with his uh M double C A R T N E Y. I'm pretty accessible on there and in sort of yeah, business, social, network sense. People, if they're like a, you know, someone's actually in the C V D frequently in Sydney, I'll often visit like, you know, some of me again, the other day I was like in the city with a barrister. So it's I'll sometimes come into the city where I have time to have chats personally with people. So they prefer a more in-person style of conversation and learning uh about like product. And um, yeah, in terms of my email, it's sort of like will McCartney at you know habeas.ai, which is H A B E A S.ai. At a more basic level, people can just go check out the product. So they can book in a demo call if they go to the website itself and they just look up habeas. Um, they can either then book in a demo for themselves or the firm, have a chat regarding partnerships or whatever it may be as well. Or they there's also for the sole practitioners out there, a lot of them just go straight to the source or you know, they'll just go and uh get their free trial set up or get a subscription set up because they are confident already from a colleague or something. And yeah, so there's plenty of ways to engage with us and our team. Um, and I think you know, beyond that, if I missed anything, I I'm not sure. But yeah, it's kind of like very easily accessible through a number of modes. And there's probably quite a lot of reading people can also do related to habeas through the blogs we publish, a lot of the resources that have been published by other providers, uh, referencing habeas and what makes it stand out. And and that's a lot of like this sort of this knowledge stuff. And I think finally, as well, what we try and do is like if someone is going to work with us, um, ensuring there is that onboarding context, or we can we can also provide services related to deeper onboarding, deeper AI upskilling and stuff like that, which is helpful because if firms are looking for a more white glove or sort of um guided way of getting into AI that can be part of the longer-term initiative.

SPEAKER_01 1:58:35

Yeah, totally. No, that's great. And you've got interactive guides, which are awesome. You can kind of see how to use the platform and deep dive into those specific features, and you can also book a demo, which I saw on your website. So um, thank you so much, Will. It's great to chat to you. I love the product and what you're building. It's really, really good and has a great use case for lawyers and people in the legal space. And thank you for coming on the show today.

SPEAKER_00 1:58:56

Yeah, thank you so much. And I really appreciate the the consideration and just uh reach out to me and yeah, really impressed by you know what you're burning here with the podcast as well and all the different interesting people you're speaking to. So I hope hope it continues to go well. Thank you, Will. Thanks.