Mary Technology Wants to Solve Litigation's 'Fact Chaos' Problem

If you do any litigation, you know the problem. eDiscovery tools have gotten really good at taking thousands or even millions of pages of documents and narrowing them down. But what happens after that? The painstaking work of actually reading through those documents, pulling out the key facts, organizing them and building a full picture of your case has remained a largely manual, time-intensive process. My guest today calls it fact chaos. He is Daniel Lord Doyle, co-founder and CEO of Mary Technology, an Australian legal tech startup that has built what it describes as the first fact management system for litigation. The company just announced a 7 million Australian dollar seed round led by OIF Vendors and is expanding into the United States with a new San Francisco office and the launch of a self-service platform for smaller law firms. What's interesting about Mary's approach is that rather than relying on the embedding and vector-based methods that many AI platforms use, it extracts every individual fact from a document set and builds what Daniel calls a fact object enriched with metadata and linked back to its source so that lawyers always have a verifiable chain from their work product back to the evidence. I wrote about the company earlier this month on my blog at lawnext.com and I'm looking forward to hearing more directly from Daniel about what they're building and why they think this is the next frontier in litigation technology. I'm Bob Ambroggio and this is Law Next, the podcast that features the innovators and entrepreneurs who are driving what's next in law.

Before we get to that interview, please take this moment to hear about the sponsors whose generosity keeps this podcast rolling. You're looking to maximize every billable hour? Bill for Time, the most trusted name in legal time tracking and billing software, puts you back in control of your practice. Streamline your workflow with multiple timers for precise tracking, speed through billing with customizable invoice templates, and get paid 70% faster with built-in payment processing. Visit billfortime.com today to see how you can practice more and manage less. That's Bill, the number four, Time.com.

Now let's get to today's conversation. Daniel, welcome to Law Next. Thank you very much. We're just chatting a little bit before we started recording about the fact that we've never actually met before. So we're literally meeting for the first time here just a few moments ago. And I'm looking forward to having the conversation and hearing more about what you're up to here. As we are speaking, your company is announcing some big news, a major funding round, a $7 million funding round, your expansion into the United States with opening of a San Francisco office and launch of a self-service platform, I guess, for smaller law firms. Before we get to any of that, why don't we start with having you kind of just tell us a little bit about what you guys do?

Yeah, sure. So Mary Technology, I'm by the way, my name is Daniel. I'm the CEO and co-founder of Mary and we're a fact management system, which we think is a necessary platform for any litigation firm or team dealing with investigations, anywhere where there's two sides, I suppose. And we focus on extracting the facts from massive amounts of messy disorganized documents and actually sort of walking through the same process that human teams do. That is organizing those documents, utilizing those facts and being able to review them very quickly. Really with our main aim being to try and produce the fastest time to confidence and so that those facts can be verified downstream when you're actually trying to produce a work product from those facts. That's what Mary does.

This fact management system is the, Yeah, I mean, what's the story behind it? You know, there's sort of four co-founders. How did you all identify this as a problem that you want to tackle together? What were you doing at the time?

Yeah, well, Mary's actually named after my co-founder Rowan's quite famous barrister auntie named Mary, who actually helped him get into the legal world as a lawyer. And we say that she's ruthlessly efficient and so everybody deserves a Mary in their firm. But Rowan was facing what we call fact chaos, which is this concept of there is a lot of administrative work that goes into trying to to just understand the documents and understand the facts within them and how they are pertinent to the matter. And he was doing this work at the exact same time as the advent of GPT-2 or 2.5. And he actually put into his WhatsApp group that was filled with other junior lawyers and attorneys and said, I think AI is going to be huge for this type of work. It's going to really revolutionize how we do this. There was about 15 people in the group, nobody responded. That was how little known AI was at the time. In fact, he got a single love heart emoji. But then he spoke to my co-founder, Harry, who he's known since he was very young, who's a very much a go-getter. And they very quickly built a prototype really to try and assess the market and see whether or not this was actually a problem that needed to be solved and that people would actually invest in. And they very quickly got 10, 15 customers. Only small customers at the time. And then as a software engineer, they sort of bought the platform to me and I could see the value in it. I think it's also a noble problem trying to find the truth within documents. It's also a very difficult technical problem. And so that really excited me. And so me and Luke came on board. I've had a couple of businesses before, not quite in the legal space, but I'm a software engineer. And so I represent the technical portion of the team. Quite a well-balanced team because we've got Rowan who's a lawyer, Luke who's an excellent product person, Harry who's sales and marketing and me who's a software engineer. So together we sort of represent a very strong core unit that sort of covers the bases as what's required to build a very good firm, I suppose.

Yeah, you were founded, you're an Australian founded company. You're now expanding to the US. Are you in other markets as well? Where are your customers now?

Yeah, so we started in Australia, grew very quickly there. We have one customer in the UK who I can't name, but is one of the top 10 largest firms in the world, but they're going through a pilot process at the moment. But they're our only UK customer at the moment. We're very much focused on the US where we have a burgeoning customer base and we're very much focused on now, seeing if we can't replicate the success we've had in Australia in the US markets. At least at these early signs, it feels like the problem is identical and the way that we're solving it is quite refreshing in comparison to what's currently available.

Yeah, I've really been struck by the emergence of kind of a whole new breed of litigation focused tools over the last few years driven by generative AI. It's my perception and I've covered legal tech for a long time has has been that the kind of the litigation sector hasn't really been all that well served by technology that that I mean, there's been sort of, you know, trial management, trial presentation tools, that kind of thing, discovery tools. But very few tools that kind of go to the heart of the kind of problem you're trying to get at. You know, I know you're a programmer, not a lawyer, but I mean, when you were kind of creating this product and developing this product, did you kind of look at the market and see gaps there in terms of what was being, what was available in this area?

We didn't start out as a tool that extracts facts. We came to that as an insight, really, in that we actually tried to leverage AI and other emergent technology to try and support the sorts of laws that Rowan understood, which was litigative and dispute resolution type law. When we started doing it, what we recognized, and I think this is being recognized by a lot of people at the moment, AI is useful. But if you're going to use it to produce something in a world, in the legal world, where there is such a low fault tolerance and where the risk is just not balanced, if you get something wrong here, and this is in a very messy set of documents, if you get something wrong, you're on the line for it. And you're also sort of helping trying to understand and solve matters that really are super important to people at the personal law level, but also of course at the commercial level. I mean, these are very high value matters. So you can't just rely on AI and say, Hey, give me this answer. And when it comes back to you with an incredibly polished authoritative answer, it's ultimately inference. And if you were to ask it, Hey, how did you come up with this answer? It doesn't go back and review how it's done that work. It just gives you another answer that sounds very polished and authoritative that suggests how it might have done it. And so we recognize that that was a major problem from the very beginning. And so we really had to try and understand what was the atomic unit. What is the correct unit that you have to build the remainder of your product with? And that thing we decided was the fact.

Yeah. And you talk about you talk earlier and I you're talking a lot of your website and elsewhere on the sort of concept of fact chaos that that lawyers now face. I mean, can you can you talk a bit more about sort of how how this has been done versus how you enable it to be done?

Yeah. Ultimately, this is a human manual process at the moment. A matter enters a firm once it's been assessed. And even that assessment often has a large portion of this work in it as well for a lot of firms. That's where you will, let's imagine we start at the point where you receive documents. You might receive tens of thousands, if not hundreds of thousands of pages of documents. And all of the value is within there somewhere, but you have to try and understand exactly what you want to prove out, what are the legal elements that you actually need to find evidence for. And to do that, you have to often follow the rules of your firm in taking these big documents, splitting them up, naming and classifying the documents. And this is a huge manual job often done by teams of junior lawyers. Once you've done that, you'll read through every single page and you'll try and extract the facts that are most necessary for your particular issue that you're trying to understand. So we call that the idea of custom chronologies. Like I'm trying to understand exactly what happened with this person to do with this particular issue, for example. The way that's done is humans will sit in a room and read through it and it'll take an incredibly long time. And often they'll have time constraints and they won't necessarily be able to build all of that time out. And so there's a heap of issues that sort of mount up here, but I'm sure almost all of your listeners recognize that in litigation, still the massive majority of the time, even with all of these advances in e-discovery and things like that. The majority of the time is spent on this discovery and document review portion of the work. So ultimately, a senior lawyer will then review the chronology, push back on what they might need to understand more. And it's from that base of facts that have been extracted or understood, or the documents have been understood, that they'll begin producing whatever is the first piece of work product that's required. And that's different, depending on the sort of matter that you're dealing with.

With Mary, we really begin at the life cycle of a matter and stay with you throughout until the conclusion of it. As soon as documents arrive, lots of our customers use us for the assessment phase, you will put everything in there. So it might be that you're going to pull it down from Relativity or push it in from iManage or NetDocs. And we'll accept all of those documents that you think are relevant for us to understand. We'll begin to construct an overview and a summary of the matter. We'll also do all of that organizational work for you. The stuff that I believe computers can actually automate. So I suppose maybe to begin this, should actually tell you, I think there's no way to automate this process fully from an AI perspective. I think all human judgment will always be required, but there are some that you can actually, well, I tell you, I actually have to sit down with quite a lot of lawyers who do get very concerned about this. And just from the inside, I can tell you now there is a huge challenge, maybe an impossible challenge for AI to ever sort of do the sort of judgment based thinking in law that people expect it will. Law is, ambiguity is a feature of the law. It's not a bug. And how you actually understand the facts and which facts are important and what you look for and all of that sort of stuff is always going to require critical judgment by lawyers, especially how you present it eventually. Ultimately, we'll organize all of those documents and that is a part of the job that I think is fully automatable. So we understand what a firm's naming conventions are, how they split their documents, how they categorize them, and we do that for them. We, of course, this is where a key element of this comes in for me, which is productive friction. I think, and this sort of backs up that point that I was just saying, even if you've done this perfectly, a lawyer doesn't care because they still have to review it. So even if you say that you've gone through and done all of this work on these documents, they still have to review it. How do they do that currently? Well, when they've outsourced this document work to an external team, they'll spot check it. So for example, we enable them to spot check how we've split those documents. That's an example where we actually don't allow you to move past certain points until you've done the verification work so that when you get to the work product, Mary can tell you, hey, out of these 300 facts that you've leveraged in this particular document, 250 of them have been verified. You need to do additional work on these 50 before, from a human perspective, before we can move forward from an AI perspective and actually leverage this document and the facts inside of it. So, sorry, just to continue, we split all the documents and do all of the work that your team's normally doing manually. And then beyond that point, we then have extracted all of the facts, every single fact. This is why it's quite different from say a large, a unified interface firm like Harvey or Legora, who ultimately store your documents in vaults and they store them in the most efficient way possible, effectively through embeddings, which I know I get a bit technical sometimes, Bob, so you might have to tell me to shut up. Ultimately, they have to try and be able to answer massive questions over massive amounts of documents. And in order to do that in the most efficient way, they turn them into embeddings and store them in vector stores. That's a problem because in litigative law or dispute resolutions, it's the nuance that these compression machines remove in order to store efficiently that really matters to lawyers. And so we have taken an entirely different approach. And so we extract every single fact and then attempt to consolidate, enrich, almost apply metadata. You can almost imagine like an object of a fact where we've built that fact over time. Who's involved in it? Are there any conflicts or inconsistencies? Where is it based? What issues is this related to? That sort of thing. And then we allow lawyers to effectively cut, slice and source the matter at those three levels, facts, documents, and the summary. And they can ask questions, but when they ask questions, that's where we bring in some of that productive friction so that what they really understand is what I'm looking at, something I can be confident in. So I'll stop there for a minute just because I've been going on for a little while. I'm just trying to, partly I'm just trying to visualize this. So are they working with all of this within your platform? Are they working on this at all within the documents that they're working on? What's the sort of workflow around?

Maybe I'll give you an example of the last portion there. We have a portion of the platform that's called Fact Explorer. It looks a little bit more like your standard, like I can ask a question, but also we have tried to understand the matter. And so we've predicted some of the use cases that you might want to do at this point. Again, similar to where you see the use cases underneath ChatGPT or something like that. So when you ask a question, what pops up is a fully cited version of the answer that you needed, along with the ability to click on that citation and it actually opens up the document to the right as well as the fact at the bottom so that you can actually go and assess exactly where that fact comes from. If you want to, you can go back to that fact within the, cause there's almost three sections. There's the summary, the document index, which you can explore in different ways and then the facts and then there's the fact explorer. So, yeah, but I should say, Bob, one of the things that I think is very different about Mary is how easy it is to use. Which is why we're releasing the self-serve because what we found is that a lot of our customers were actually sharing it with other firms and their clients. And when they were doing so, they weren't necessarily requiring training or anything because we really lead with a value. Innovation without complexity. Even the most technophobic lawyer should be able to come onto Mary and enjoy doing this work.

We'll be right back with more from Daniel Lord Doyle of Mirri Technology. When we return, we'll hear about what sets Mirri apart from the big platforms, the company's move into the US market, and why Daniel thinks the last mile of litigation is where the real opportunity lies. Stay with us.

Is your law firm tracking time and expenses but struggling to see the full financial picture? Bill for Time's new Advanced Customizable Reports solves this by bringing all your productivity and financial data together in one place. Effortlessly customize reports to see what matters most and spot unbilled work before it impacts your firm's revenue. No more piecing together spreadsheets or guessing where your firm stands. With Bill for Time, you get the insights you need to run a more profitable practice. Learn more at billfortime.com. That's Bill, the number four, time.com.

Back to Law Next. I'm Bob Ambroci and I'm speaking with Daniel Lord Doyle, co-founder and CEO of Mary Technology, which has built a fact management platform for litigation teams. Before the break, we talked about the problem of fact chaos and why Daniel believes that even the best AI tools can't be trusted without a verification layer. Now let's get into how Mary is bringing that vision to the US market. Thank you. But yes, it's in. Got a lot of good phrases like productive friction, innovation without complexity. I like this. Ah, I like the concepts.

Well, I think you might like the idea behind productive friction because there was this amazing study that came out of Wharton recently that sort of vindicates a lot of the work that we're doing. It's actually by two researchers called Shaw and Nave. And they basically did the old cognitive mirror test. But this time, what they did is they were going to give people a set of questions. For 50% of the people, they were going to give them an AI tool that works and gives an accurate answer. For the other 50%, they were going to give them an AI tool, but they secretly told the tool to give them slightly incorrect answers. And so what they were trying to test is, will people rely on AI even if the AI is giving them incorrect answers? Now, unfortunately, the results were pretty worrying in that I'll have to double fact check myself here, but it was 88% of people relied on it. And they even felt more confident when they were using it. Now they split the test down three ways. I know it wasn't 88%. It was 88% of people who felt more confident. was 56% of people would still leverage it when there's still a false result. But they kept making the test more and more interesting. So the second one was time pressure. And of course more people used AI and more people got the wrong answer. But the third one's the most important for me. And that was, they then said to everybody, If you get the question right, we'll pay you and we'll tell you after each question whether or not you got it right or wrong. And there was almost a concerning amount of still misuse of the tool in that people would still rely on the AI output. And I think it comes down to this concept of there's always been two types of thinking, gut type thinking, but Daniel Kahneman's thinking fast and slow. There's guts. There's judgment based like critical thought, which lawyers are just incredible at. That's what you're being paid for. But now there's this third artificial thinking, which is actually outside of us and is voluntary to use, but we're all getting very used to using it. And when you're using it in it, where there is this risk asymmetry, if you use it in the wrong way, you get a terrible result. You're actually putting your client at risk and yourself at risk. And so we think at Mary that this productive friction is necessary so that lawyers, junior lawyers and senior, can practice that type two thinking, their judgment, where it matters most without having to focus on all the parts where it doesn't necessarily matter, which I don't think AI has been so good at so far.

The end result is still, you're not replacing the lawyers, but you are saving them a lot of time, right?

Well, think time, for me, this first wave of AI, it's all been about time. I think, of course, that's very critical for certain areas of law, particularly those who are based on contingent-based pricing or no win, no fee type thing. Got a little crush released. Yeah, yeah. And I think that's super important for those people. Time is the greatest factor for them because they want to do more volume work. But I certainly think that there's other vectors even for them. And that is, can I actually get paid for all of the work that I do? A lot of the work that I currently do, I can't charge for because I'm just repeating work or I'm getting multiple people to do the same job. But mean, more it's about, can we find facts that you might not have found previously? When you want to review what you need to review, can you do it quickly, but also effectively? Can that work be shared efficiently? And so I think this document review process, you've probably seen that there's been a huge surge of people working on it because it's such a massive bottleneck. Probably the biggest in litigation.

Yeah. And that's kind of what I meant earlier when I was talking about the fact that that's a part of the litigation market that just hadn't been well served up until, up until January.

Yeah. Think it might have been impossible because the one thing that I would love to sort of share with people who don't understand the space as well as maybe I do, is given that I'm entirely obsessed with it, I doubt there's many people, but they, the documents in commercial sort of the, what's the correct word? I mean, if, like contracts, for example, are very organized documents. And when you're dealing with M&A, for example, the majority of the documents are contracts and they're so organized and they are defined by proxy. Everything you need for a large language model to understand that contract, as well as whether or not it refers to another one, has to be inside of the document. Otherwise it doesn't constitute a contract. It's a poor job of being a contract. Whereas the alternative isn't that in litigation work, what you're looking at is massive amounts of, for example, medical documents, whether it's handwriting. Historical abuse documents that are ancient yellowed paper and cursive writing that's difficult for people to read, numbers, sheets, times, all of this stuff. It's so many different formats and you have to be able to understand and predict what documents are needed, be able to extract that information in a useful way and be able to sort of explain what the negative space says. And there's a lot of documents. So you might get, you know, whole ream of docs where only 10% of it is really relevant for the matter.

Yeah. I mean, eDiscovery software tried to get at that problem to an extent. I you talked about Relativity before. I mean, using predictive coding or whatever else, the ability to kind of at least sort of identify relevant, non-relevant, frivolous, that sort of thing, but not to the kind of granular detail you're talking.

I'm still, I'm still very much in the belief that actually the biggest time saving will remain what that those e-discovery tools have done. I mean, they take a million documents and turn them into 50,000 for example. You know, it's impossible to understand how much time that's probably saving people. And they've done so in a way where the entire industry can trust it. They know that there's a small error rate, but it is very small. And I think that eDiscovery tools have done an excellent job. I don't think that necessarily they're the best tools to do that last mile work where you are trying to understand the facts of those 10,000 documents, for example, and that's where Mary sort of lives.

Yeah, eDiscovery, I actually used to work for eDiscovery a couple of years ago. And one of the things I was well aware of is that lawyers were slow to adopt eDiscovery technology. I mean, it took a long time for lawyers to start to get comfortable with, especially with using AI and eDiscovery. You know, even still, even still, there are a lot of lawyers that just got to do it the old fashioned way. Not quite not quite file boxes in warehouses, but still very uncomfortable with allowing e-discovery with allowing any kind of technology to make sort of calls about documents, what are you seeing with your product? What kind of adoption have you seen? What kinds of obstacles do you see yourselves as having to overcome to get wider adoption?

Well, look, we've gone on a journey where the adoption has been rapid. We are yet to go from customer to sort of rather a pilot to what we call a proof of value and onto an actual customer because and you know, when these pilots are asked, you know, does anybody want to take part in this? Everyone's hand goes up because I think I don't know if this is maybe good to say, but I think this is part of the work that people don't enjoy. And they find it very, very time consuming and obstructive for them actually learning how to do legal work. A lot of the work here is copying and pasting. What we want to try and do is advance people to that judgment work as fast as possible. I think the biggest barrier for us, it's a question. We haven't necessarily seen adoption challenges, but I think the one thing that we need to do and continue to do better than everybody else is this idea of productive friction and trying to help people become confident quickly. I think we've gone through a wave of AI here where it's almost seemed like magic. And I think we need to strip back the magic and really insert lawyers where they should be inserted and help them do those things that they really should be thinking about as much as possible. And I think doing that properly will mean that we grow, continue to grow at the same rate as we have. I think we're one of the firms that have been specializing on that.

So we're talking now as you've announced a $7 million funding round. I don't know whether you had any previous raises.

Eh We've had a smaller pre-seed. This is our seed round and we're very happy to be invested in by one of the best VC firms in Australia, which is called OIF. They've had some excellent successes that you probably might have heard of, but Empress Capital and Sydney Angels as well, they've been wonderful supporters throughout. We only added one additional investor in this seed round. And yeah, I mean, we really want to leverage this to expand into the U.S. and we've already started doing that. So really great signs so far.

And why is that important to you? Is that just simply because that's where the largest market is?

Well, I think we do this because we want to have the maximal impact. And there are 16,000 firms in Australia, 550,000 or something in the U.S. It makes a lot of sense. But also I think it's fair enough to suggest that if we're going to be the category leader in fact management, we need to understand how this work is done in the U.S. and in the UK like we're doing with that larger firm in London. And ultimately all of Europe, we will start with like the, all of globally, but we're starting with the common law countries who have a similar process around document review.

Yeah. Yeah. Are you targeting, and obviously you're targeting litigation firms, you're targeting particular types of litigation firms, plaintiffs firms, or as a defense firms, or what are you looking at?

Yeah, so we also work with some of the largest internal legal teams in Australia as well, because they actually suffer from some of the similar problems to say a plaintiff personal injury firm, where, you know, they have a very small legal team that is responsible sometimes for tens, if not hundreds of thousands of customers and staff. And they aren't necessarily rewarded for some of those internal investigations in terms of getting the best result always. Sometimes it is about actually just trying to get through some of this work and understand what to do next. So internal legal teams have been very, very helpful for us as well. So internal legal teams, commercial litigation, family law, personal injury. We've got a lot of focuses, ultimately wherever facts are required. That's where we help best.

Yeah. Part of this announcement is the launch of this self-service capability that you talked about earlier. Again, think that you're, again, based on the materials I've been providing, that's kind of focused more on the smaller firm end of the market to allow them to do that. I mean, can you talk about a little bit about what's involved in that? How does that work? How does somebody get up and running on it?

Yeah, sure. The reason we did that is because the majority of our time we do spend with the larger commercial teams and the larger enterprises. And we recognize that we had a lot of people who wanted to come and use Mary, who we simply just didn't have the time to dedicate to them, which is good because we've actually tried to have a first time user experience that is incredibly intuitive. And people were able to actually come on and work on it themselves. So we actually have lent into that because one of the biggest pieces of feedback we get is, wow, this is very easy to use and it's very intuitive and you know, we love it. So the best way to sort of do that is to actually go and head to the Mary website and sign up now, at which point you'll actually get a hundred free page credits. And so you'll be able to assess whether or not Mary is useful for you. I mean, we're very value led. So by all means, we'd love you to come on and actually test some of your documents. All of the data is, we used to work working with the largest firms with the most complex security requirements. So you can be rest assured that you can visit our trust center on the site as well to understand how we look after your data. But yes, absolutely. Head on there, get your 100 free credits and test a small set of documents in there and see how Mary works for you.

So get there, plunk on a credit card, upload some documents, and start to play around a little bit? Is that it?

Yeah, you'll be yes, it's that easy. It will just say, create your first matter. And so you click that button and it will guide you from there. Um, but obviously with a hundred pages, you might not get too far. Um, but obviously the concept is that you can actually see how Mary works. It's what works best for us in demos, Bob. Like we, we don't bother with the demo. That's I think most people do where they've got their perfect set of data. The bit that we love is actually. Give us the hardest thing that you've got and let's show you how it actually works on the demo. So I think we're just sort of taking that learning and putting it into self-serve.

Yeah. You know, there are, I mean, there has been a lot of growth of litigation focused legal technology platforms over the last few years. Who do you kind of view as your key competitors and how do you distinguish yourselves from them?

I mean, interestingly, I think that at the smaller side of the market, I think our biggest competitor is ChatGPT. Mm-hmm. Yeah, I mean, that's what we're sort of hearing and seeing. There are, of course, other people who are focused on this type of work. I mean, Relativity has Relativity AI. But there are other small things, but it's a pretty blue ocean. The sort of work that we're doing is not. It's not. Done by the majority of firms because it's a very different approach. This concept of actually trying to extract every single fact and being able to leverage them to sort of produce this verification and review experience is quite unique. I would suggest that there is a firm called Wexler who do focus on the upper side of the market who I respect. I think they've got a really good outfit. Um, yeah.

Why wouldn't I just use ChatGPT then?

There's a number of reasons why you wouldn't use it, but I'll give you one example. You're working with five team members to do this work, and it's a very, very large document set. And you want to obviously not have that document leave your secure environment. So you probably want integration when you put all of the documents in. There's quite a strong limit on the number of documents you can put in there. In terms of number of pages, number of docs. When you do do it and you get that answer out, whatever that answer is, let's imagine it's a chronology or it's quite a difficult legal question, you're not going to have a fact database and a document organization tool in there. And so you're to have to consistently prompt and ask questions. And again, when you ask it how it's formulated that answer, it will be very difficult for you to understand whether or not a human has done any of that verification, e.g. you in this case, because you can't share that particular chat and have a consistent piece of work that's always stable. For example, auditing is incredibly important in this type of work. If you are working with a colleague and you update your fact or delete a document or delete a fact, you really want to understand that that's happened. Or if you come to a conclusion and produce a work product, you're going to want to know, okay, which facts have actually been relied on here? Because again, if you ask ChatGPT how it's come to that answer, it just produces another authoritative answer on how it's done it. Whereas we have an complete tracing, we call it, which is the concept of we need to be able to tell the lawyer if they want this work product, here's what they can rely on. Here's what they may be missing. Here's what else they need to put in. I mean, there's lots, Bob, but I mean, ChatGPT is not an appropriate tool for this type of work. Particularly given that some matters last years.

So I'm curious, because I just left San Francisco. You're opening an office in San Francisco, or you've already opened it, I think. Oh, why San Francisco? What made you choose that location within the United States as your headquarters?

Yeah, well, there's actually a bit of a logistical thing here in that San Francisco and Australia share a better time zone relationship than say New York. But I certainly think San Francisco is just the beginning. Yeah, of course, you know, California has a massive legal market. It's obviously where lots of talent is as well. So, yeah, it's just the beginning.

What's, as you're coming into the US market, what's the most important thing you want this market to know about your company?

Number one, that we're here and what we do and for that to be clear and obvious. And I think one of my challenges is that the language I use is probably Australian at the moment in terms of the way that I describe things. I think we need to try and learn more. I think the big thing is if you have a deep and frequent problem that involves document review and understanding the facts so that you can actually feel confident about using AI to produce work product, we're a really good company to talk to. And we're super keen to learn from you. I think that's the main thing. We're not a 10,000 person, 50 year old firm. We are very keen to jump in, understand what you need and actually build for that. Yeah, I suppose the one thing I wanted to know is that we're very, very keen to talk to you and understand what you need. We speak Australian here, so you can get by, okay, I think. I suppose I mean like things like saying attorney rather than lawyer or you know, barristers don't exist in the, yeah, no barrister. I'm sure your lawyer is pretty interchangeable here. That's the easiest question. That's good. Yeah. But what would you say? Would you say matter or case?

Well, that's also used pretty interchangeably, but I think for litigation, I tend to say case for litigation, but that's a matter of taste probably more than anything. I think of matters as more, I don't know why I think of them. Well, no, I guess I was going to say more transactional, but that's not even right. But a lot of the practice management systems refer to them as matters, as a generic term.

What do you think, given you know a little bit more about our firm, what do you think is the key thing that I should be talking to US firms about?

Ah, you know, I think, well, I think that that last mile point you made earlier in the discussion, I don't I honestly don't think there has been much in the way of technology that addresses that drop off from where the e-discovery platforms drop off and from where the litigation starts. That's really where you need to get in. You you've as you say, you've narrowed down the collection of documents to collect and you really need to be able to start to dive into the facts. And again, you mentioned Relativity AI's trying to sort of increasingly getting into that space to some extent. Everlaw with its deep dive kind of lets you do a little bit of that fact exploration, but not again, not with the I haven't I haven't tried or even seen your product, but just what I'm hearing you say and what I've read about it. None of those are doing it with the kind of the nuance that you're that you're able to provide. So I think that's it. I think it's a little bit of educating. Probably a little bit of educating the market, although I do suspect that for, as you suggested earlier, that for a lot of litigators, this is a pain point they know very well, and it probably won't be a hard sell to get them to understand what it is you're offering.

Well, it's certainly been wonderful so far because the product sort of does the speaking for us. So, you know, I've been doing lots of calls quite late at night. So is the rest of the team. Because I've been flying all around at the moment and the same problem exists. A lot of the tools that they've tried to use, they get what looks like a perfect output, but they still have to do 100% of the manual verification themselves removing. How beneficial that product is for them. And that's, I suppose, the gap that we're trying to close.

Yeah. All right. Well, any, final words, anything else you wanted to say that we haven't had a chance to talk about her?

Oh, we'll be at New York Legal Week. Just hanging out or your exhibiting? Yeah. And there'll be five of our team there. So, uh, the other three co-founders, Luke, Harry and Rowan, well as our chief commercial officer. Um, so please do visit if you are there and have a look at the platform. Um, what we're showing off there is really more around what we're doing for the larger commercial litigation and general litigation teams. So that's probably the audience who's there. So yeah, we'd love anybody who is going to that to go and have a look. And of course, if you are a law firm and this is a deep and frequent problem, you actually want to produce verified output. Please head to merrietechnology.com and give it a try.

Dan, it's been a pleasure talking to you and pleasure learning more about what you're doing. And welcome to the United States. You're in London right now, I think you said, but you'll be here soon.

That's right. That's right. Pleasure to see you, Bob.

Alright. Thanks for joining us for today's show. Hope you enjoyed it. If you'd like to share your own thoughts or comments, please do so by messaging me on LinkedIn or email me directly at ambroji.gmail.com. Law Next is a production of Law Next Media. I'm your host, Bob Ambroji. I hope you join us again next time for another episode of Law Next.