The AI Pivot That Took BlueJ to 300% Growth

But I also had this thought, which was pivotal and led to the founding of Blue Jay, which was I just imagined myself in the year 2040 standing at the front of a classroom at the University of Toronto at the law school with chalk in my hand thinking, I saw this coming 25 years ago and I probably should have done something a little bit differently than what I'm doing. I'm, I'm still teaching the same kinds of classes in the same traditional way. And in the interim, everything has changed—legal education, in the law—and I feel like I really, I imagined my future self feeling profound regret about not having done something differently.

Welcome to the Startup CEO Show. I'm your host, Mark MacLeod. In this episode, I sit down with Ben Allary, co-founder and CEO of BlueJay Legal. Being a professor is a full-time job. Being CEO of a growth-stage startup is a full-time job. Ben is doing both. His company is going through 300% year-on-year growth as they work towards their vision of creating what Ben calls the legal singularity. They are focused on tax. It's a massive market. $29 trillion in tax is collected every year. $500 billion is spent administering that. This is Ben's TAM. I hope you enjoy the episode. Ben, welcome to the Startup CEO Show. It's a true pleasure to have you here. How are you doing today?

I'm doing great, Mark.

Thanks for having me. Yeah, I, you know, we were chatting a little bit before and I've wanted to have you on the show for some time. A, you know, like I'm a recovering CPA, did public practice for 6 years, hated tax, and you know, you would've made my tax life a lot easier if you existed back then in the dark days. And then also I'm fascinated about someone who is simultaneously a professor and a startup CEO. So maybe just to set context, could you tell folks what BlueJay does?

Sure. BlueJay is a tax research platform that leverages AI to make tax research nearly painless and immediate. And so you can take basically any tax research question and run it through BlueJay and get an answer that is accurate, comprehensive, reliable, and linked to the underlying legal authority that informs the answer. And so it's essentially, Mark, tantamount to magic. And it's like this thing that I wish existed when I was a tax law student back in the early 2000s at the law school at the University of Toronto, and then even as an academic for the first 10 years of of my time as a tax academic. I spent a lot of time doing tax research, hundreds, thousands of hours doing tax research. And this is— That sounds deeply painful. This is a dream to be able to simply articulate in plain language a tax research question and offload so much of the underlying labor of finding the right sources, reading them, synthesizing them, and producing a coherent on point, backed up, cite, you know, cite citations of the sources where you can track where you can trace where the answer is coming from. It's just, it's kind of a dream solution. So that's, that's what BlueJay, that's BlueJay's kind of value proposition is, is, you know, we will just take the pain out of producing better tax answers. Got it. I love that.

That's fascinating. There's a lot that I want to explore there, but we'll come maybe back to the platform and your vision later on. I'm curious about the origin of this kind of on two dimensions. First of all, the name. We're recording this while the Toronto Blue Jays are in the playoffs for the first time in a decade. Where does the name Blue Jay come from?

So it doesn't, it doesn't come at all from the baseball team. I didn't think so. It comes from Yeah. I remember clearly where I was when the name occurred to me. I was walking down Charles Street. I was westbound on Charles Street, just walking past the McKinsey offices there. Oh yeah. Beautiful office. Victoria campus. Exactly. And it kind of, it struck me that Blue Jay would be a really compelling name for the startup that, that I was planning with my my colleagues from the University of Toronto from the law school. And the reason why it was so compelling is in legal writing, if a J appears behind someone's name, that means they're a judge. So it'd be like McLeod J would be Justice McLeod and that you'd just be McLeod J. So Blue Jay is like Justice Blue. And then the question is, okay, why the color blue? Where does blue come from? Well, it comes from the University of Toronto having the Varsity Blues, and it comes from IBM being Big Blue. And we thought at the time, and so if you cast your mind back to 2014, 2015, IBM was putting a lot of energy behind promoting Watson. And so we thought we'd be relying on, at least in part, Watson technology, IBM's technology, in building Blue Jay. And so it was like, it was this very tight, tight series of characters, 5 characters, BlueJay, that captures the fact that this is a legal research solution. It has its origins with the University of Toronto. It has this connection to IBM with respect to the underlying technology. And, you know, we could get the domain name, which is also an important consideration when you're starting out. And also, I guess, maybe secondarily, you know, Blue Jay references an animal. And for whatever reason, we humans have very good, you know, recollection of animal names and bird names. Rather than naming it some, you know, name that has no connection to anything, just something, some random assortment of syllables, Blue Jay is something that people can remember, not least because it also has this association to the city of Toronto. And we were also founded in the city of Toronto, and Toronto famously has the the Toronto Blue Jays as our baseball team. But that was at best like a tertiary consideration. The primary was capturing the relationship between IBM and University of Toronto and this being a legal technology solution. Secondarily, about Blue Jay being like something that would be quite memorable. And then this tertiary thing, which hasn't at all proven to be a problem. Some people question out of the gate, like, "Will the baseball team be irritated that you're using this name?" I said, "Well, I guess we'll find out." But it hasn't at all been a problem. And nobody's going to confuse our tax research offering with a professional sports team. So, so far there's been no market confusion about that whatsoever. That's great.

I love it. This episode is brought to you by Limitless.

Limitless is an exclusive community of VC-backed CEOs who want to scale themselves and their companies without limits. The biggest outcomes in the startup world are founder-led. Meta, Amazon, Shopify, Airbnb, Canva, NVIDIA. Behind every one of these companies is a founding CEO who didn't just scale their business, they scaled themselves. They added one zero at a time and today run companies with billions of dollars in revenue each. That's the goal of Limitless, to give you the tools, strategies, mindset, and community to lead at the highest level without sacrificing your health, relationships, or purpose. Through group calls, an elite peer network, a proprietary AI coaching bot, members-only library, and in-person experiences, Limitless gives its members everything they need to keep scaling themselves and their companies for as long as they want. Go to Limitless.ceo to learn more. I'm fascinated.

You know, I would imagine it's hard enough to become a professor in a top-tier university, and I know that it is very hard to be the CEO of a venture-backed company. You're doing both at the same time. Tell me about that.

Well, what would you like to know, Mark? I guess exactly.

Like, first of all, like, are you a masochist? You know, do you sleep? The skill sets for one feel different than the other. So like, how did you— I understood, I understand how you became a professor. You've studied a body of knowledge at, you know, various levels and have mastery over your domain. I can see that how that would give you the, the impetus for the idea for this, because you have domain knowledge, very unique domain knowledge and see a gap. But then I guess, I guess what I'm most interested in is how you, we we can talk about the simultaneous thing after, but how did you transform yourself to also become a venture-backed CEO? Yeah, it's, uh, you're right.

Uh, your question implicates a whole bunch of things around, you know, personal identity and professional identity as well. Right. So to become a legal academic, uh, requires you kind of have to go all in on that. On that identity in order to do the work for years and years, right, in order to build up that expertise. And, and so you're absolutely right. So, you know, as an undergraduate, I studied economics, finance, philosophy. That brought me to being very interested in law and economics. And so I found myself at the University of Toronto at the law school, concurrently doing a law degree and graduate work in economics. And I was still extremely fascinated. That took me to Yale Law School, and I did graduate work in law at Yale. And that took me to the Supreme Court of Canada where I was a judicial law clerk to Madame Justice Louise Arbour. And then I was very interested in becoming an academic, and the stars aligned and the University of Toronto had an opening, and I interviewed at a number of other law schools and was lucky enough to have a number of options. But I decided to return to the University of Toronto. And at the time I was, it was 2004, I was 26, I think maybe the youngest tenure-track law prof. Yeah, I would imagine.

Similar age to you. In the law school's history.

Yeah. And, but there was a need for tax faculty and that was the area that I was really most keen on. And so it just, the timing worked out extremely well. It was my dream job and I was so nervous, Mark. I was so like, I was like panicked. I'm like, I'm younger than a lot of these students and very motivated to do an excellent job. And so I put my head down and wrote dozens of articles. I co-authored with my mentor, David Duff, a senior faculty member at the law school, several editions of Canadian Income Tax Law, the second, third, fourth, fifth, and sixth editions of this. 1500-page book that provides a full introduction to the rudiments of Canadian income tax law. And it was through that process doing thousands of hours of deep tax law research. And the peculiarity of having a book like that is no sooner is it published than it's out of date because there are new cases all the time coming out of the tax court, out of the Federal Court of Appeal, out of the Supreme Court of Canada. Parliament is constantly every year tinkering with the Income Tax Act and changing different components of it, and then the CRA is coming out with new rulings, technical interpretations all the time. It's a very dynamic system, and so you have to keep updating the materials every couple of years. But as soon as that book is published, it's a guarantee that it's at least out of date in several particulars. In the book and it really gnawed on me, the laborious nature of updating that book, but also the fact that it was always out of date. And so when I was teaching out of the book, it was like, "Oh, this thing has changed and this thing has changed and this thing has changed." And students are like, "Just make it stop." Like, just— You're like, "Talk to the government, dude." Yeah. But your question is really about, so how did I go from this path, which obviously requires being all in and taking tax law as a subject of study extraordinarily seriously and really dedicating my life to the life of the mind and learning and teaching tax law to starting a venture-backed tech company. And I think the precipitating event really was when I became associate dean of the law school in 2011, and I approached the Dean. I said, if, if I'm going to, if I'm going to take on this, this job, one of the things I would like to do as part of this job is revisit the, the way that we deliver a legal education here at the University of Toronto. I want to lead a curriculum committee and see what we can do to, you know, basically revamp how we're approaching a legal education. I had just been through it a decade before. As a student, and then I'd been teaching in the curriculum for the past several years, and I felt like there was a way that we could do it to improve it materially. And I had a senior colleague approach me and say, Ben, are you sure you really want to take this on? It hasn't really been done since the early '70s. And, you know, I'd caution you against doing it. There are some colleagues who felt very strongly about the last curricular reform in the early '70s and, and the relationship was never really the same after they had, you know, pretty personalized disagreement about what to do about the curriculum. And I didn't take away from that caution what I think my, you know, hoping to be helpful senior colleague was trying to get me to do, which was like, maybe I should just spend my time doing more research, academic research, rather than trying to improve the curriculum. Instead, what I heard was it hasn't been done since the '70s, and now it's the early 2010s. And so it's been 40 years since somebody did this. And then that got me thinking, okay, so first of all, it's probably, you know, long overdue that somebody really sit down and try to figure out how to improve the curriculum because it's starting to really show its age. But the second thing I thought was, well, if it's been 40 years and someone like me showed up to do this, maybe it'll be another 40 years before someone like me shows up to do it again. Mm-hmm. And then that triggered the thought in my mind, which was, well, what's going to change over the next 40 years that we should be taking into account as we think about how to revamp the curriculum? And then my mind started swimming, Mark. It was, okay, well, you know, everybody's familiar with Moore's Law that, you know, computing power is growing exponentially. The cost is coming down exponentially. I was aware of the work of Geoff Hinton happening over in the computer science department and with his industry work. And I was aware that, you know, almost all of legal materials were becoming natively digital. There's a huge industry of digitizing a lot of the traditional materials, taking them from the stacks in the, in the libraries and translating them into digital format. But also like everything new is digital natively in law. And so if you have those ingredients, computing power, algorithmic improvements, digital data, the content is all potentially, you know, analyzable using machine learning and AI. But oh, we've got all these ingredients. And, and that got me thinking, okay, this has big implications for what we do with the curricular reform, but even more so, it has big implications for the legal profession. It has big implications for the judiciary. And I kind of felt like there's like a freight train coming straight at the law school with AI on the front of that freight train. Coming for us. And I was like, okay, so this is going to be a thing. This is for sure going to be a thing. And then I started thinking, thinking about, well, you know, we'll, okay, we'll take this into account with the curricular reform. We will. And you know, this is, it eventually triggered a whole bunch of like a change in my scholarly research agenda as an academic. And I started refocusing on law and technology, but I also had this thought, which was, which was pivotal and led to the founding of Blue Jay, which was I just imagined myself in the year 2040 standing at the front of a classroom at the University of Toronto at the law school with chalk in my hand thinking, I saw this coming 25 years ago and I probably should have done something a little bit differently than what I'm doing. I'm still teaching the same kinds of classes in the same traditional way. And in the interim, everything has changed. In legal education, in the law. And I feel like I really, I imagined my future self feeling profound regret about not having done something differently. And so then I, then I challenged myself, like, what is it that I should do differently? Because I've just spent the last, you know, X number of years, you know, full on in building an academic reputation for myself, a body of work, a scholarly research agenda. And I couldn't shake this thought that, well, somebody is going to leverage AI to produce something that's essentially a magical tax research engine, and that's going to be a huge thing for the law. And I thought, and this is where maybe, you know, it's academic naivete kind of kicked in. I'm like, Well, maybe, maybe that person should be me. Like, why not? Like, I can probably do this. Like, I know, how hard can it be to do a tech startup? It's like, it's like the most naive thing you could possibly think. But, you know, oftentimes, you know, it's— I think, I think this is— I'm not the only founder with this, you know, with this thought. It's like, oh, like, I don't know, people do this all the time. You know, it can probably I could probably do it too. And so, and then I thought, well, if it's not me, who's it going to be? I'm like, well, it's not going to be a tax professional, an accountant or a lawyer who's bought into a partnership and has a, you know, a steady stream of work coming from longtime clients. It's probably not going to be a math or computer science grad who doesn't understand tax law, probably isn't really interested in tax law or the problem that I'm interested in. And even if they were interested, it'd probably take them a decade to catch up to where my understanding is and the amount of like scar tissue and pain that I've accumulated in doing this kind of work for this long. And I thought, well, maybe I should do it. And so I was having lunch, you know, with, with, you know, my, my colleagues, a number of my colleagues, and two in particular were very interested in this, two brilliant colleagues, Anthony Niblett and Albert Yoon. Anthony Niblett grew up in Australia, and, and did a business degree and a law degree at the University of Melbourne and then moved to Cambridge, Massachusetts, did a PhD in economics at Harvard, taught at the University of Chicago for a number of years, and then moved to the University of Toronto. And Albert Yoon, who grew up in upstate New York, went to Yale College and then Stanford Law, Stanford PhD, taught at Northwestern for a number of years before joining the faculty at the University of Toronto, where you know, good friends. A lot of our scholarship used the same kind of quantitative methods. And, and it's like, guys, we'd been talking about this for months. I'm like, I think, I think I really want to do something. I think we should, we should do something. I don't want to feel that regret of not doing anything. And I think it's, it's a like, it's a really good bet. If nothing else, we'll have, you know, a very interesting adventure trying to do this. And they're like, okay, Ben, as long as you take the lead in doing this, then okay. And I said, I wouldn't have it any other way. So, okay, I'll take the lead. And thus was started Blue Jay. I think what I really didn't appreciate at the time is just how long it was going to take for the technology to develop to the point where we could produce a solution that had this quasi-magical payoff, which was you can answer any tax research question and it's going to come up with a pretty darn good answer. That took much longer than I expected. But we're riding this exponential curve and the improvements are now really, really exciting from week to week, month to month, especially year to year at this point.

Hmm. Was it a tech transfer thing? Spun out of the university or is just, this is a separate thing. I'm a professor and on the side I'm doing this.

We started, yeah, I mean, we started working on it while we were, while we didn't, we didn't have a company. And so we did reach an agreement with the university and we, the university took a small equity position in Blue Jay when we founded it and, and assigned all the IP. We collectively assigned all the IP to the company when we started it. Got it. Okay. And then, and then we operated them totally independently. We're very like careful to separate everything out and, you know, we hired all the students that Blue Jay hired, all the students on real agreements. And there's, you know, we're lawyers, right? And law professors. Yeah. So we were really good at keeping watertight compartments there.

Right. It's fascinating. So, you know, When I try to picture your reality before this, so before becoming a technology entrepreneur, you're going super deep in tax, in probably very specific areas in tax. And therefore what I'm picturing is like a lot of patience. You're going to do the canonical research, have like the understanding, take whatever time it takes to produce these documents. And you're tenure-track, so like, well, I'm just gonna do this for the rest of my life, so time's not really a big consideration. And then now you become a venture-backed CEO where time is everything, like speed is everything. How did you make that switch?

Well, I suspect I'm temperamentally not a pure academic type for one thing. So, you know, um, you know, So the first several years as an academic, it's publish or perish, right? So it's like you're on the tenure clock and it's like, okay, go, go, go. It's not really like, take your time. It's, you know, the expectation is you should be consistently producing published, peer-reviewed work year after year and of a sufficient quality and quantity that, you know, you will have developed an international reputation for your work. By the time you are up for tenure. And at the time at the University of Toronto, it was after 5 years that your tenure file would be assembled, and then the expectation would be that you'd be upgraded from assistant professor to associate professor as of the 6th anniversary of your original appointment. And so that's not a long period of time. It's like— It's true. It's true. It's go, go, go. And that's kind of my, I kind of have, I think by nature, that attitude that I do not like to waste time and I'm able to get lost in the flow of academic research. And so if I'm working on an interesting problem, time just melts away and you get lost in your thinking and reading the sources and crafting the arguments and doing the research. And I absolutely used to luxuriate in that, in finding the flow. And, um, you know, there's psychological literature about flow and finding flow in your work. And I used to regularly experience that flow in doing academic research, but it wasn't— I didn't feel like it was meticulous. Maybe somebody watching me might've been like, oh, look how patient that guy is just sitting at the desk with all these books around him doing this work. But it felt, it felt very active and very creative and sufficiently challenging and sufficiently interesting that I could just get lost in that work. And I was very, you know, those years were extremely productive as an academic leading up to the founding of the company. And then you're right, once you've started the business and you've got investors and the clock is ticking and they're looking for signs of progress and you've got the invested capital and they're looking for their internal rate of return and they're looking for signs that you're experiencing that ARR growth and with a view to eventual profitability and a really nice exit for them. Yeah, I think the truth is I'm probably neither a typical academic, nor am I probably a typical founder of a venture-backed company. I'm, I'm a little bit of a different character than both, probably a little bit of a fringe character on each, just enough to make it into that Venn diagram overlap between the two sets.

I can see both roles being— I don't want to belabor this simultaneous thing point too much. We can move on, but like, I, both roles seem all-consuming. I have no idea what it's like to be a tenure professor and a dean of a department, but I assume it's all-consuming. And I know what it's like to be a CEO of a growth-stage company. I only coach CEOs of growth-stage companies and it, it's all-consuming. And of course, if I'm your investor, I'm selfish. I, I don't want you to be distracted by anything. I want you to be all in and just like, every waking moment you're thinking about how to make me rich. So does that create any tension for you?

Not at the moment, but for the first several years, I did maintain a significant active appointment at the law school and was growing and leading Blue Jay. This year, I'm now on leave, unpaid leave from the university to focus entirely on the business because we're going through hypergrowth now. I had a moment almost precisely a year ago, Mark, like right before Thanksgiving of 2024, and I was scheduled to be teaching 3 courses in the semester beginning January 2025. And I was just like, I cannot see how I'm going to do this. I really I don't know how I'm going to do it. And even if I can do it, it's like, it's absolutely not the best use of my time to be at the front of the classroom teaching 80 students tax law. That just doesn't scale. I really need to focus on the business. And so, um, I asked the university for permission to go on unpaid leave and that was granted. I had previously scaled back my appointment to a three-quarters appointment. So I was 75%. Which meant that I could teach 2 classes per year and satisfy the teaching requirement associated with that level of appointment. And I could concentrate those courses in a single semester, so I would teach from January to April each year. And January to April was always a grind because I was in the classroom several times a week as well as doing all of my Blue Jay things. But actually, worked out okay. And I think my investors were okay with it because they did see the credibility-granting aspects of holding an academic chair at a university like the University of Toronto. It's like very good kind of credibility-boosting for the business. It's like, yeah, and I would get academic speaking engagements that if I were just a tech CEO, maybe I wouldn't get the same kind of invitations to speak. And it gives— it lends academic credibility to the subject matter. And the university also, thinking about it from the university's perspective, why would the university permit someone like me to do this? Because it might be a distraction from my research and other things. I mean, part of the answer is, well, I scaled back my appointment to 75%. According to the university's rules, that allows me to spend time on outside income-earning activity so long as I disclose what it is that I'm doing and the magnitude of those activities, which I did. And so it was okay from— it was like, again, it was like kind of barely okay from both perspectives, right? It was like, it was barely acceptable to the university. Of course, the university also had an economic interest in the success of the business because they had a small equity percentage of the business. But also that scaled back my compensation from the university. And so I was being paid less, but it was also good for kind of the innovation story that the University of Toronto really believes in, I think, and, you know, wants to promote entrepreneurship and commercialization of research. And so that was also quite appealing, I think, certainly to, you know, the deans who served as leaders of the law faculty while I was there, and also I think to the senior administration at the university. So it was all like just kind of like threading the needle. Like I'm barely the kind of guy who could be both a tax academic and venture-backed CEO and I'm, I could barely figure out the overlap between the university's interests and, you know, the venture capital investor interests, but was able to make that work at least until this year. And then this year, like, I give up, like, I can't, I can't do that anymore.

Yeah. Yeah. That's fascinating. So, you know, you're not a, you were not a product person. You may be one now. You were not like a computer software engineer, but you are customer number one, right? You probably designed this for yourself, and I'm guessing a lot of people watching or listening kind of are in those shoes, like, hey, I know what needs to be built. I've got the domain knowledge, but I'm not a technologist. I'm not a product manager. How did you work with those people and kind of, well, you're, you need to manage them. You need to know how to hire great people and then have them turn your vision into a product? Just tell me about that. I got extremely lucky.

One of my co-founders, I mentioned him earlier, Anthony Niblett, I mentioned he grew up in Australia. He actually moved to Melbourne when he was about 8 years old and around the same— from England, he was living in England with his family. His family relocated to Melbourne, Australia at about the same time, Mark. Another kid named Brett Jansen, who had been living in South Africa, moved with his family from South Africa also to Melbourne. And so happens the same street in Melbourne, and they were the same age and they were the new kids in class together and they became fast friends. They're both kind of outsiders and both wickedly smart kids. And so they kind of fell in together. And they remained friends. Brett went to the University of Melbourne, studied computer science, and was living in Melbourne at the time that Anthony and Albert and I co-founded Blue Jay. And Anthony was Skyping with Brett and just said, oh, you know, by the way, I'm starting a tech company. And Brett was like, what are you talking about, Anthony? You don't know anything about, you know, software or technology. Like, what are you doing? You're an academic. Anthony said, "No, no, yeah, you know, with Ben and Albert and we're doing this thing and Ben's into tax law. He thinks we can build something that can answer any tax research question." And Brett's like, "Okay, this is really interesting. You know, I have to learn more. I need to talk to Ben." And so Anthony brokered an introduction between Brett and I over Skype and we were Skyping. And I kind of described what we were up to and Brett's like, you, what are you doing on the tech side? And it's like, well, yeah, that's one of the problems that we have. We need to figure out who's actually going to lead the software development. And Brett said, well, you know, I'm a software architect. I'm like really interested in what you're doing. And I'm interested in relocating to Toronto as well. And so we were able, it took several months and I got some friends some help from a friend of mine who was running at that point his own immigration law firm, and he agreed to help us out pro bono to help do the work to get Brett a visa to come and relocate and start working on Blue Jay with us. So we founded the business in April of 2015, and it was early October 2015 by the time Brett was actually landing in Canada. I went to Pearson and picked him up with Anthony. And actually just this past weekend, Brett celebrated his 10th anniversary here at Blue Jay. That's cool. And the rest is history. And we consider the 4 of us to be co-founders, Brett, Albert, Anthony, and myself, because it really was indispensable to get a proper technical lead. You can't do what we've done without somebody who really understands how to build the software. And so I really credit Brett with figuring out, okay, how do we actually approach this? How do we build a product that's compelling? How do we translate, Ben, your vision into shippable code, production quality code, and actually develop it. And then of course it's been a team effort, right? We have more than 100 people now, and over the years, you know, so many people have contributed to pulling this together. We've got a great set of developers, a fantastic set of data scientists, a product team that is really speaking to users and advisory groups. And like, it really is, it takes on a life of its own once you once you cobble together the initial elements. And so, I mean, you trace it all back and it comes down to like, I'm really glad that Brett and Anthony moved to the same street in Melbourne. Yeah. How about that? Because that's been really, it turns out like a very important thing for BlueJeans success.

What a synchronicity. I love that story. Thank you. Maybe let's talk about enterprise sales for a minute. You know, I have several enterprise sales companies and, you know, like I tell them all the time, hey, if you sell to like, I don't know, the Fortune 500, well, you got a TAM of 500 customers and it's a bit of a land grab, kind of whoever gets there first, you know, they're going to get most of the market value in that category. And you don't only sell to accounting firms, but like in the accounting vertical, there isn't a Big 500 or Fortune 500. There's like a Big 4. It's like super concentrated. And I know there's lots of regional firms. I, back when I was practicing, I worked for regional firms, but like, The big four are really big. And, and, um, I'd be curious to know, like, how do you go about navigating that and kind of just penetrating when it's so concentrated?

Well, I'll tell you, it's not easy. Um, yeah, it's not easy. Uh, we've been fortunate. I think maybe this circles back actually to one of the benefits of having that deep academic background is it's possible for us to have conversations with those kinds of firms on, you know, trading on, you know, our academic background in tax law. And so very early on, Mark, we had investment from two of the Big Four accounting firms within— Cool. And they participated in our pre-seed round and our seed round. Oh, wow, that's early. And it's like, It's actually like a very unusual thing for Big Four firms to do. But, you know, could go in and say, this is the vision of what we want to do. Here are some early prototypes. Here's what we think the system is going to be able to do. We'd be very interested in co-developing this and, you know, making it like basically dogfooding this with you and seeing, if what we're building is appealing. And so getting feedback from them early on and aligned feedback, I mean, they're investing. And it was really, really helpful because then of course they're also inclined to be the first paying customers after you've built something that works. And then it naturally turns into, okay, well, we should roll this out to some actual users and let's get serious about this. And And they did become paid subscribers to it. And then once you have those kind of reference customers, then it becomes easier to talk to other potential customers in the market. And it's kind of a signaling thing to the rest of the market. Oh, you know, even the Big Four are interested in using these tools. And so I guess, I don't know, I guess the magic there is, well, how did that come to be that we were able to get those Big Four to invest. And there, I think it's finding people within those organizations who are willing to champion these ideas and who really buy into the vision and were willing to put up some of their own reputational capital in support of the idea and go to bat for it. And I think that comes down probably to really good storytelling and a mix of storytelling, but also you know, just having that connection to academia that I think really bolstered the story.

And don't forget a good dollop of luck.

Oh, 100%.

I'm curious, like, is there any dissonance with your value proposition and the economic model of an accounting firm? Meaning like, I sell billable hours, presumably before you I was, all those hours where I'm spending looking for precedents and blah, blah, blah, like I'm billing that and now it just shows up. So like, was there any concern on the part of your customers with that?

Sure. I think there are a few different responses, but none that totally overcome the objection. Right. So one response is yes, but you're writing off a lot of that time because a lot of that time spent inefficiently. So you end up. Writing off a bunch of that time. And that's time you're not going to get back. You're paying your folks to actually spend that time to hunt down the answer, but you can't actually pass the cost of that time onto your clients. And so there's some loss there, irrecoverable time. I guess another dimension of an answer is, well, there is this better way that exists and you're not actually doing all the research that you could possibly be doing, especially during busy season when there's just far too much work to do. And so it's just like, let's just keep going. And so actually there potentially the quality of the research that you're doing just suffers because there's just not enough time in the day. Of course. And so you're kind of, you're just limited by the sheer number of hours in the day and you're going to, you're still going to build that time, but maybe you can elevate the quality of the work product in the same amount of time and you're not going to actually sacrifice any billable time. Now I guess the third thing is, you know, there's the fact that this technology exists now, and especially now it's definitely at least 10 times better than the other ways of doing tax research. The fact that this exists now, it's just like you can't really countenance doing it the old way. People don't want to spend their time doing it the old way, and it's a great way to lose people if your competitors are using the new technology. You're either going to lose your employees or you're going to lose clients because the clients aren't also going to want to pay you to do all of this work. If another firm is using the better way of doing things and is charging, you know, maybe half as much as you're charging for the same work product, and probably your work product isn't quite as good in terms of quality, and they're paying more for it, the market's going to figure it out. So, you know, the workforce is going to figure it out. If you think about the employee market. And then also the ultimate end market, the clients are gonna figure it out. And so I think there's market discipline on the firms that introduces to them adopting it. But you're right, there is that kind of that friction that has to be overcome. Where we don't encounter the friction is, for example, with the end clients. So the big Fortune 500 companies you were talking about earlier, they all have like internal tax groups and they are like the customers and they're like, oh, this is fantastic. We can now like massively improve the throughput of the tax folks that we have internally and we can better discern those questions that are things that we really need to go to outside professional advisors to help us figure out versus, oh, this is good enough. I'm sophisticated enough in my understanding of tax using this, this kind of platform, I can find the answer. I see how it's supported and that's good. We're good to go. We can ship the answer. I'm, I'm comfortable with my professional opinion on this one and they can just forge ahead and they don't have any of the same qualms. In fact, they're like, this is saving us a lot of, a lot of spend. I totally— and then, and then think about the dynamics, right? Because if they're using the tool in those big enterprise environments, And then they're getting answers from the professional advisors that aren't as good as the answers that they're able to produce internally immediately. They're like, what's going on? How come it took you a week to figure this? We figured it out in, you know, 15 minutes using BlueJay. And that's an embarrassing place for the big firms to be.

I love that. So does it reach a point where you're just getting pulled in? Like it's just sort of expected that, you would be installed?

Yeah, we're going through hypergrowth right now. We're growing over 300% year over year. We're adding 10 new logos a day, basically every day. It's like this is a wild, fun time as we're, as we're kind of, you know, growing. Hmm. I love that. So congrats. We're getting— yeah, that, you know, the net revenue retention is 150% plus, it's landing and expanding. New logo is amazing. Gross revenue retention's over 99%. It's just like, it's kind of like, this is, this is wild fun right now.

You're a big guy, but you must need to hire bodyguards to fend off all the investors.

They must be just clamoring after you. Yeah, there's a lot of interest. Um, yeah, there's a lot of interest. Champagne problems. Mm-hmm.

Um, Maybe last big topic, let's talk about LLMs and, and, you know, how they relate to you. You know, when you started 10 years ago, they probably, I'm guessing, were not on your radar screen. Maybe Geoffrey Hinton and other thought leaders at the university might have given you a preview of the future, but my guess is, is like you started without that. And so how has the presence and the momentum of of LLMs either helped or hindered, you know, you and your vision for your company?

Oh, large language models have been hugely instrumental. I really like, I really would characterize Blue Jays history as having kind of two, two parts so far. There's kind of version one, which was 2015 through to the end of 2022, and then everything that's happened from 2023 to now. And so when we started in 2015, you're right, like large language models weren't really a thing. Of course Watson was like using natural language processing, but it wasn't as efficacious at all as what would come later. And so what we started with was supervised machine learning and we were collecting bodies of case law on particular tax research topics and then making predictions about how courts would resolve different kinds of fact patterns, different kinds of scenarios based on those decided cases. And technically, I think a lot of the work that we did was incredibly powerful and groundbreaking, and it's still the best way to answer a lot of those kinds of questions. But the problem was it couldn't answer every tax research question you had because you needed a body of case law on a particular kind of topic in order to do all of the work to build that model. And then those models, you know, require care and feeding and maintenance. As new cases are decided, you need to maintain that set of cases, and then you need to update the model. There's a lot of work that doesn't scale all that well and results in like an uneven user experience. Because if you don't have some— if you don't have a model on a tax research question that somebody has, it's a terrible user experience. They log into BlueJay, and we're talking about version 1 of BlueJay. They would log into BlueJay, they're like, oh, I loved how it like helped me figure out whether worker should be treated as an employee or independent contractor, I wonder if it can do this thing where I'm wondering if it's X or not X, and they would log in and it's like, oh, BlueJay doesn't do X or not X. And so, you know, you have a couple experiences like that and you're like, I just, I can't be bothered to look, I'm just gonna, I'm just gonna do my conventional analysis on this. And so we were well aware of the limitation of, you know, of not having complete coverage in BlueJay with the version 1 product. So early in the fall of 2022, OpenAI released into the OpenAI Playground, uh, the DaVinci-003 model of GPT-3. And I was playing around, we were monitoring the progress on large language models and I was playing around with it. And I remember at the time being so excited about the capabilities of that DaVinci-003 model. And I, I think if you would go into on LinkedIn and find some of my posts of around that time, I was posting about, you know, just how performant DaVinci 3 was and how like it was, people weren't noticing how great it was. And so I was very excited. And then of course ChatGPT got released at the end of November, and I think even OpenAI were surprised by just how quickly and excitedly things took off from there. But it got to the point where at the end of 2022, I was like, okay, I see clearly now that we should be pivoting. And, you know, we'd grown to several million dollars of ARR with our kind of V1 approach to Blue Jay. But I'm like, this isn't going to scale to where we need this to be. We should just put all of the V1 business in maintenance mode and let's make this next jump, which is to seeing if we can leverage large language models to answer any tax research question somebody has. And so we spent the first half— I committed the company to the first half of 2023. Let's see what we can build in U.S. federal income tax law to answer any U.S. federal income tax law question. And kind of the rest is— the rest is history. We went from, you know, a couple million dollars of ARR by the end of 2023 to just shy of 9 at the end of '24. Now we're 3/4 of the way through 2025 and we're approaching $30 million in ARR. And it's just like, it's just been pretty fun. And Q4 traditionally has been the strongest quarter for this business, so.

That's amazing. You know, when I think about like the wide variety of things I use large language models for, And then I compare that, like, you know, maybe researching every possible thing. Like tax, I'm wondering if this is true, like tax is, it's large, but it's finite and it's described already. Like, first of all, there's an Income Tax Act. It's large. I love, I read the Income Tax Act back in university and I love the preamble about how income tax was a temporary measure in Canada. But anyway, let's set that aside. But like, it's, it's large, but it's finite. Uh, and I guess, is there less room for hallucination and interpretation? And it's more just like finding the right fact? Like, does that make it more, uh, I guess leverageable by AI? I'm not sure if that question makes sense, but yeah, it just struck me as a finite body for AI to get its head around.

Maybe, and, and maybe I'd put it a slightly different way, Mark, but I think it takes us to the same place, which is there's a known universe of kind of authoritative materials that you would have to have your arms around in order to produce a tax research opinion. If you were a tax accountant or a tax lawyer, there is like a known universe of things that you would look at in order to, you know, produce your tax analysis. And there are things that you, you just like cannot cite to that they have no legal authority, they're irrelevant. And so it would be things like the tax legislation itself. You mentioned the Income Tax Act, of course, the Internal Revenue Code in the United States, similar. And there are other, you know, there's other legislation provincially at the state level in the US. And then you've got, you know, all of the case law. So the courts are producing case law adjudicating tax disputes between governments and taxpayers all the time. Then you've got tax administration, the Canada Revenue Agency, the Internal Revenue Service in the UK, HMRC. Who are issuing rulings all of the time to taxpayers, interpreting the legislation, interpreting the case law. And then you have academic commentary, you've got professional commentary, but you get— you can describe all of that stuff. And then when you get there, you're like, okay, anything else? Not really. I guess maybe some, some commentary that, you know, you can find on the web, but the most valuable there would be classified as professional commentary, like a law firm or an accounting firm might have some verbiage on their websites talking about different aspects of tax law. And then you're kind of there. You're like, that's the referenceable universe for answering tax research questions. And it is tractable and you can analyze it using retrieval augmented generation. There are certain ways in which you go through like the logical exercise of producing a tax research analysis and it's describable, understandable, tax people know it when they see it. And if it's making reference to the right kinds of sources, not including stuff that has been superseded or replaced by subsequent developments, it's like, okay, that's a pretty good analysis. So I think so. I think there's a lot of power in choosing a well-defined vertical and saying we're going to produce, we're going to focus on producing the best possible tax analysis and we're going to leave a lot of the other possible things that you could harness this technology for to others to really focus on. We're just going to be the absolute best solution on the market for this big enough area. And maybe just one word on like the total addressable market you mentioned earlier, like if you're selling to Fortune 500 companies, it's like that's your universe. I think people often forget that tax is huge. This is how we finance governments worldwide. Every year, $25 trillion of tax is paid, and the administration and compliance costs of running the tax systems is about 2% of that $25 trillion, which is— if my math is right, that's about like half, half a trillion dollars, right? And so this is like $500 billion potential market of real expenditure associated with improving the efficiency of tax compliance and tax administration. And so there's lots of room for an improved method to actually provide a huge amount of savings and value in the world. So it may sound niche, like, oh, tax research, but it's actually, I think it's quite big.

I had no appreciation. I mean, I certainly know that everyone pays tax, but I had no appreciation. For the size of that TAM. That's wild. So, you know, you're an overnight success, 10 years in the making. What's your vision for the next 10 years?

I'm really in pursuit of something that as an academic I've described as the legal singularity. I published a book with University of Toronto Press with a fantastic colleague, Abdi Adid, back in 2023 called The Legal Singularity: How Artificial Intelligence Can Make Law Radically Better. And the vision is really that the law should be transparent, it should be understandable. We should have universal legal literacy. We should be able to leverage algorithms and AI to provide absolute clarity to the law everywhere and on demand. And tax law is a very good place to start because of the size of the market and because of that, that the power coming from focusing on that vertical. But there's no reason in principle why it can't be used across all of law to improve our social operating systems. And so I think that's really— it's kind of mission-driven, is how do we actually make law better leveraging AI? And so I see the next 10 years being a continuation of this pursuit, Mark. And so I'm not on this podcast to sell my book, but I would recommend if people are curious about hearing more about this vision to look up The Legal Singularity. It's on Amazon, it's on Audible. And you might find it interesting. Very cool.

Well, Ben, I had a sense you were up to big things, but this kind of, as a CPA, it kind of blows my mind how big this is and how big it could be. So I'm cheering you on from the sidelines and I can only imagine how busy you are. So I really appreciate you making time to come on the show.

Thanks for having me on. It's been a pleasure.

Awesome. Hey, thanks for listening to the Startup CEO Show. If you'd like to connect with me, me, be sure to visit my website at markmacleod.me or follow me on LinkedIn @TheMarkMacLeod or Twitter @MarkMacLeod_. And if you wanna tune in again next week, be sure to subscribe on YouTube, Spotify, Apple, or wherever you get your podcasts. We'll see you next time.