S14E7: Building the AI Layer for Legal Work with Scott Stevenson, Co-Founder and CEO of Spellbook

One legal bill took half the cash out of our bank account, and I was like, oh, what's going on? Like, this, this is horrible. And immediately my brain just started obsessing over this problem. Like, how many entrepreneurs are starting companies and then just not able to afford to do the basics to protect their IP, to raise some, a small amount of investment, or even opening a coffee shop? There's all this paperwork and red tape that seemed extraordinarily expensive to me as a young founder, and I got really obsessed with that problem.

Welcome back to The Room Podcast Season 14. Hi, Mads.

Hi, Clauds.

So excited to be here.

So excited to be talking about everything that's going on in 2026 in tech.

There's so much to discuss.

Each season, we sit down with some of the most thoughtful people on the cutting edge of the startup ecosystem to talk about what it really takes to build and scale enduring companies.

This season, we're diving into the ideas, inflection points, and experiences that have shaped some of today's most impactful companies. And the people behind them.

From Nextdoor to the Ring founder, we are diving into everything from early product decisions to growing at rapid scale.

These conversations offer a window into how great companies are actually built.

And if you enjoy the conversations, which I think you will, make sure to subscribe so you never miss an episode.

And if you wanna stay connected to everything we're building at The Room, including community events like our annual Inside Summit, you can subscribe to our newsletter at theroompodcast.com.

Thank you so much for being a part of the community and welcome to season 4. Of The Room podcast. Rippling is one of those tools I wish I had even earlier when I was building my startup, Pryve. If you're building a startup, every minute matters, but it's surprisingly easy to lose hours to onboarding paperwork, fixing payroll issues, or setting up laptops for new employees. We've ran into this constantly, and it's not hard work, but it's constant and it adds up. That's where Rippling comes in. Rippling is a unified platform that lets startups run HR, payroll, IT, and finance in one system from day one. Rippling's startup stack replaces disconnected tools that don't sync now with a fully connected platform. It lets you get back to the real reason you started your company. Right now, venture-backed startups can get 6 months of Rippling's startup stack for free. Head to rippling.com/theroom and sign up today. That's R-I-P-P-L-I-N-G. Rippling.com/theroom to sign up for 6 months free today. Focus on what you're building, leave the rest to Rippling. rippling.com/theroom.

Hawkins Cooley supports the most innovative entrepreneurs and investors in fast-moving and high-growth sectors, addressing their myriad of legal needs. But the firm doesn't just provide end-to-end legal and business counseling to its startup clients. It also facilitates introductions to key advisors and sources of capital. Perkins Coie's interactive website, Startup Percolator, offers access to programs, resources, and rich dynamic content designed to assist entrepreneurs on their startup journey. To learn more, go to startuppercolator.com and perkinscoie.com.

Legal tech continues to be a very compelling market in which AI is creating radical disruption. Earlier this season, we heard from the Crosby founder about how AI is reshaping how legal teams operate from the inside of businesses. But today we sit down with Scott Stevenson, the CEO and co-founder of Spellbook, to take us to the other side of that equation, how tools are being built for the lawyers themselves. Scott's path to founder market fit is anything but conventional. He grew up on a rugged island outside of Newfoundland, Canada, where isolation breeds creativity, and his outsider's perspective ultimately became a competitive edge. He studied computer science immersing himself both in the philosophy and electronic music ecosystem as well, and ultimately stumbled into what he's building today through a personal pain point of having a huge legal bill that almost wiped out his prior startup's company. Since founding Spellbook in 2017, Scott has raised over $80 million, which includes a $50 million Series B at a $350 million valuation. He is currently building for the future of lawyers, as we discussed, and has over 4,000 customers across 80 countries with hero customers such such as Nestlé and Dropbox. In today's episode, key insights and themes include navigating a pivot at scale, how Scott's personal belief and metaphysical idealism have shaped who he is as a founder, and ultimately why radical transparency in contract data is about to change the way the world makes deals. Let's open the door.

Scott, welcome to the room and thank you so much for joining us today.

Great to be here. Thanks for having me.

We're delighted to have you. So we'd love to start at the beginning as we do with all of our guests. So please tell us, where did you grow up and how has that shaped your view of the world?

I grew up in St. John's, Newfoundland. So it is an island on the far east coast of Canada, one of the furthest east locations you could go in North America and the furthest place you could possibly drive from Silicon Valley, literally, you know, distanced from tech. But I was always very plugged in online through Twitter and online message boards and things like that, developer communities. I was always building things growing up, very tapped into what was happening in the Bay Area as well. It's a magical, very rugged, beautiful island. So it has that sort of, I don't know, creative island culture, similar to like Iceland or somewhere like that, where it's a little bit isolated. There's a lot of creative people who have to make fun for themselves. Like a lot of artists and musicians would not tour to Newfoundland if they went coast to coast. There's so many musicians in Newfoundland. I grew up making music like many of my friends, and the whole music scene is mostly local. One of my favorite activities that I do pretty much every day when I'm in Newfoundland is walk through the woods and just think. It's a great place to be like independent-minded, creative, and to build sort of an outsider's view on the world, which I think can be helpful in sort of coming up with contrary perspectives that might have alpha in them and really inspired me.

I love this alternate view of island lifestyle. Definitely the opposite of a beach and a margarita, but honestly sounds incredibly soothing and a little remnant of growing up in Seattle. We had the San Juan Islands, which similar were just very out in nature and rugged and had just access to so many beautiful things you can see and do. I love how that shaped you and informed where you thought about your passions and your interests. So in that vein of early identity, Did you always think that you were going to become a founder?

I always considered myself a creative first, and making things that are meaningful in the world is what I cared about most. I first was interested in making video games, then I was interested in making music, and then finally I was like, oh, you know, building a startup is another kind of creative act. And actually it's a very big, complex, chewy creative act that takes, you know, like a decade or more to really do well. That really compelled me. I was always interested in tech, but I think as like a, a creative at heart.

I love that. I personally also would say I'm a creative. Having that creative mindset, people don't often associate with being technical, but decisions you make in people's codebases and where it's gonna intersect with people's needs is creative. So, you did go on to study computer science at Memorial University, and I saw that you used to describe yourself, maybe perhaps still do, as a metaphysical idealist, which is a bit of a philosophical view for those who don't know, that believe reality is fundamentally mental and not physical. Could you unpack this a little bit for us and just relate more about your undergrad experience?

Besides tech and creativity, one of my other passions has been philosophy, and I see philosophy as this pursuit of truth. And my, my blog is called Competitive Philosophy. And what actually ended up really drawing me into startups is this idea that what we're doing is trying to figure out what is true before competition. And if you can do that repeatedly, in small ways and big ways, you create an enormous amount of value. And I think if you're like a contrarian person who likes thinking about the world and what's real and what's true and what's not, and what are people deluded about and what aren't they, it's just an amazing place to be because you actually get an objective score through startups on like, are you good at figuring out what's true about the world or not? Is really what startups tell you and what investing tells you. What I love about investing in startups besides the creative element is this pursuit of truth and figuring out what's true about the world before others do and actually getting to get an objective score back on was I right or was I wrong? That to me is like applied philosophy. I'm very, very interested in philosophy on big and small issues. I love metaphysics. I've thought about metaphysics for a long time. What is going on? Why are we here? Why are we conscious? I love these big questions in the same way I love the startup questions. And I always say my most contrary big bet about the world is that physicalism is not true and metaphysical idealism or something like it is true. And we could have a whole podcast episode on this. Idealism is not a new idea, and there's like the new— there's like the new agey, you know, hippie version of this stuff. There's this guy Bernardo Kastrup who wrote some amazing books. He's an engineer, and he had this really modern take on metaphysical idealism. His first book in the series is called Why Materialism Is Baloney, and he just puts together this amazing case for why our view of reality is like being made out of matter, primarily out of matter, and you build up that matter and you connect it together and you amibas and then you get consciousness at some point. He makes an amazing case for why that's obviously not true. I won't sum up like the whole book now, but I'll give a brief argument for why I think this is so compelling. There's two big problems we haven't figured out. One is, what is consciousness and where does it come from? This is the hard problem of consciousness. How philosophers think about it, consciousness has nothing to do with your thoughts or do you have a concept of self necessarily. It really boils down to one thing. It comes down to you have a point of view. And the question you ask to say if something is conscious is, does this coffee cup have a point of view in the world or not? And most of us would say, probably not. So it's probably not conscious because it doesn't have a point of view. We've never figured out why, like, at what point does matter start to have a point of view? If all of this matter out there and all of these rocks and grass and helium atoms, we have all these things out there that don't have a point of view. At what point do they suddenly have a point of view? And, um, there is no one in the world who has ever been able to answer that question well, in my opinion. So that's one problem that we've had. And then the other problem is Stephen Hawking has this quote where he's like, we have all these equations that describe how the world moves, but we don't have an explanation for what breathes fire into the equations. Like what, what is actually making those equations move what's around us and what's creating a world for the equations to describe. And I won't make a good full argument, but if you flip things upside down and you say, consciousness exists first, there is perspective, and that's the one thing you know exists is that you have a perspective. That's the one thing that is indisputable, is that you have this first-person perspective of the world. And if you say that's the thing that we're going to assume exists and then we're going to explain how everything else comes to be, that's actually a much, much easier tractable problem. In fact, like every night when you close your eyes and dream, you see that happen. You have this sort of perspective and you see how reality comes to exist. And I'm not saying you can just will reality into existence or anything like that, but I think, you know, a really cool analogy is like Bitcoin, for instance. Like, in Bitcoin, there's no objective reality. In Bitcoin, you have these Bitcoin nodes, thousands of nodes around the world. That's like us. And Bitcoin does not exist outside of the perspective of one of those nodes. You could have a node in China behind a firewall that has a different view of reality than a node in North America It's the nodes themselves that kind of enable everything to exist. And I think if you flip things around and take this like distributed computing analogy, it actually ends up being a much easier way to explain what's going on. So that's a little bit of a, a deep rabbit hole.

First of all, it's such a relevant rabbit hole, right? I think we're all facing this existential question at the moment of how consciousness will manifest in agentic experiences, especially when it's given yourself, your identity, your digital assets that can create a manifestation of who you are. And then why does it decide to say one thing versus the other? We still don't totally know. And so there's just this opportunity right now to redefine what's real. But as you said, that might be a slightly other podcast.

Sure. Yeah.

Yeah.

But I think you're right. Like it is an extremely relevant question right now because Eventually you have to face, as these AI models get better and better, it's like, is this thing conscious? Like, multiple studies that show like if you're rude to AI models, they perform better and they are more accurate. So this raises this ethical question, like, should I be mean? We deliver AI to lawyers. Should we be mean to the models so that they do a better job than lawyers?

They're trained on lawyer data, right? Which lawyers have been known to be a little mean sometimes.

That's true. Yeah.

I kid.

No, not at all.

You're right though. And on the flip side, when you tell the ChatGPT that If there's a million-dollar reward, it does better too. And so that means it has a sense of incentives, which is also really interesting. So I totally agree. And ultimately this intersection of philosophy and computer engineering and where we're going to go in this new world order is like really evolving under our feet. It's a really exciting space for all of us to be building in today in 2026. But when you first became a founder for yourself, it was many years ago. Your first company was in a different space, Moon. Would love to tie in, of course, these passions around how you think about the world and your views of the world. In with this interest in music and creativity and how that birthed your first company.

Music was sort of like my second creative passion after making video games. I just, I love making things. I'm really into electronic music. I used to DJ and produce music. I just love the process and the state of flow you get in doing it. Yeah, I started working with this composer and a music professor who was addressing this problem where like they were combining electronic music with these acoustic ensembles in a classical music setting. And all of the instruments that you would use to play electronic music are kind of like a DJ deck, like sitting up there and you're twiddling knobs or whatever. And the classical music audiences would just be like, I don't really understand what this electronic musician is doing. We ended up building this new instrument. What would like a, an electronic instrument that is inspired by an acoustic instrument look like so that the audience could have a better understanding of the cause and effect. And we built this thing called the Mune that has this wooden casing and can do synthesis and drum programming and all of these different things. And it's just like a very unique instrument. It was a horrible business idea. Um, it was a lot of fun. It's a very niche, niche market. It was very expensive to build. We worked with this incredible acoustic guitar maker in Newfoundland who's this guy with a workshop in his basement who just built the most beautiful guitars you've ever seen in your life. He helped us design this wooden casing. Ended up being so expensive to like mass produce something like that. But at one point in that journey, we had to set up a company, issue some shares, protect our IP, a bunch of other like legal matters. We got a little bit of investment from an angel investor who's still a mentor of mine today. And then we got this huge legal bill, at least for me at the time. I was a new grad, didn't have a lot of cash. We didn't have a lot of cash in the company. This actually literally took half the cash out of our bank account. One legal bill took half the cash out of our bank account. And I was like, oh, shit, what's going on? Like this, this is horrible. And immediately my brain just started obsessing over this. This problem. Wow, this is a way bigger problem than electronic music performance. Like how many entrepreneurs are starting companies and then just not able to afford to do the basics to protect their IP, to raise some, a small amount of investment or even opening a coffee shop. And there's all this paperwork and red tape that seemed extraordinarily expensive to me as a young founder. And I got really obsessed with that problem. That's how I got into legal tech.

I'm excited to dig a little bit deeper into what eventually became Spellbook. So let's peel back the layer there. A little bit. It's 2017. You co-found what was then Rally. Tell us a little bit more about what prompted you to say, hey, this is what I'm going to go all in on. This is what I'm passionate about. I'm going to start a company in the space.

I had definitely been bitten by the entrepreneurial bug at that point, and I realized that entrepreneurship is very hard from the musical instrument experience. So I want to do this, but if I'm going to do it, it has to be a really big problem and it has to be a problem worth dedicating like 10 years of my life to. I worked for another startup for a little while as a director of engineering leading software development. I was involved in Ethereum smart contracts very early on in the Ethereum community because I thought that might be a solution for like legal contracts. That didn't really work out. But I, what struck me about the legal problem is just like how pervasive it is. If you try to do anything in the world, start a company, hire an employee, make a sale, you're dealing with these legal contracts. And as a programmer and an engineer, the way that these contracts were drafted and reviewed just felt like 100x slower and less efficient than what it should have been. And I was like, what, what if lawyers had tools like we had as engineers that allowed us to really code and review so much faster? I think for many people, a legal bill is like the biggest bill they ever receive, and the amount of value they feel like they get from it is not always that high. I just thought this was a huge unsolved problem and that we would try to solve it. And then I met my co-founders, Daniel and Matt. In Toronto, and they were working on something very similar, and they had a, an angel investor who is interested in the space as well. So that's how it started.

For 4 years, you, Daniel, and Matt were building document templating tools, and you saw lawyers get excited about these templates. You build them out, and then they start to abandon them because they're not capturing the nuance of bespoke deals. As a founder, in the early days, I remember downloading these templates for like employment agreements or whatever agreements, and then it was like, ah, it's getting me 90 we're 90% there, but there's like this one thing that's causing this to break down. You saw that pattern, call it 100 times. How did you keep going through those years discerning product market fit?

Yeah, we, uh, we, we kept going for a very long time before getting to like a real inflection point, about 4 years. And as you mentioned, yeah, we started with contract templates. We thought, okay, like there was no generative AI or anything at that point. So we were like, okay, we structure the data in like this nice little contract template and people can standardize on these templates. Like things will be much easier. We sold that to about 100 law firms and we actually launched over 100 landing pages over 4 years with different variations on the product, different, slightly different value props. Like at one point we were selling Shopify for law firms where not only was it a template engine, but your clients could come in and be like, I need an employment agreement and kind of pay online. And we then use a template in the backend to produce like your first draft for you. Every 2 weeks we launched a new experiment and we were just maniacal about, we're not going to scale this until we have real product market fit. Our standard for product market fit was customers are pulling this out of our hands faster than we keep up with. And we know we were around like 8, 10 people for a very long time waiting for that to happen. It was tough. We had some tough conversations with our board where they're like, let's get on with it. Let's scale the company. And I'd be like, no, I don't think we have it yet. Even though we have 100 customers, like, I don't think this is quite the right solution yet. Like, customers are buying it, they're starting to use it. The actual kind of usage retention long-term doesn't feel that good. The time to value feels really long. Getting lawyers to go to, like, a separate app outside of their existing workflow was actually a really hard problem. We kept rolling the dice and trying things and learning each time. And then the generative AI moment happened, and we had this spark of how all of our learnings could fit together to deliver something that would actually change how lawyers work.

Well, I think it shows just such incredible founder excellence to be aware that something's working, but it's not working as well as it should have. And navigating a pivot once you're, you know, at 100 customers, at a team of 10, is certainly not an easy decision. Walk us through that pivot in the moment of GenAI, which birthed the current product for Spellbook.

At my startup, Priv, Rippling helped us avoid a lot of the operational drag that comes with scaling a team. With most tools, something as simple as onboarding a new hire involves multiple systems, payroll, apps, hardware, and it can take weeks, quite frankly. With Rippling, those workflows happen automatically in one place. That means fewer distractions and more time focused on building. Over 15,000 startups rely on Rippling to scale fast without adding additional ops and HR headcount. So if you or your startup wanna move as quickly as you can and focus on what really matters, like your product and your customers, you need Rippling. Right now, venture-backed startups can get 6 months of Rippling's startup stack for free. Head to rippling.com/theroom and sign up today. That's R-I-P-P-L-I-N-G.com/T-H-E-R-O-O-M to sign up for 6 months free today. Focus on what you're building, leave the rest to Rippling. Rippling.com/theroom.

The most important thing of us sticking through it was that we really believed in the problem. So for any founder, oftentimes founders hear our story and they're like, oh, should I stick it out 4 years? Like, how long should I wait for product market fit? I think the important lesson is like, we really believed in the problem and we were really passionate about the problem. If we weren't, we would have given up and we probably should have given up much, much sooner. Yeah, so moving on to the generative AI moment, we knew lawyers needed faster time to value. Contract templates are not that helpful to lawyers. Every deal ends up being so bespoke. Joke, and we needed something that lawyers could just use very quickly in their existing workflow. The aha moment for me was using GitHub Copilot, which is probably like the first really popular generative AI product before ChatGPT. I used it for coding and it was in my existing workflow. It was just autocomplete at the time. Basically, you could hit tab, you could start writing a function or something and hit tab and it would write 10 lines for you. And to me, that was that was like incredible technology because no matter what I was working on, I could hit Tab and get this amazingly smart autocomplete that would write 10 lines for me. It was just immediately just struck me like, oh, this is gonna be incredible for lawyers. 'Cause the whole company was about how writing contracts and writing code are actually very similar. They're both sort of these logical documents where if you mess one thing up, it could be a catastrophe. They're ultimately very structured. And so we immediately were like, like, let's see, what is GitHub Copilot for contracts going to look like? And we actually raced a prototype out. The board didn't know anything about it. It was not part of our plans. We raced a prototype out in about 2 weeks and we got onto the Microsoft Store in record time. And it was basically just autocomplete in Word, but for contracts. We built it on top of GPT-3. Actually, we tried a version with GPT-2 that didn't quite get there. And then with GPT-3, it was good enough to get somewhere. And then we launched a landing page and initially we were like, okay, this will just be a cool marketing splash. We'll get a bunch of leads and then we'll funnel those leads into our Rally product and that'll be great. Let's just do a fun marketing thing. Like probably not going to be a long-term product. Within a few months we had 30,000 waitlist signups. This thing just exploded in interest and all of a sudden we had a small sales team at the time. Our sales team calendar, we would do onboardings with every customer, like a 30-minute onboarding. I looked at our sales team calendar, it was every 30-minute slot for the entire week for every sales rep was blocked and they weren't even taking lunches. They're like trying to eat between meetings. So it was just like this massive influx that was just like lightning in a bottle moment for us. And that's when I went to our board and I was like, I've been telling you we didn't have product-market fit yet. I was like, I think now we have it. We should probably change the name of the company and pivot in this new direction. It didn't take much convincing to do that once we had We had users using it very quick. Part of the magic was we were in the lawyer's workflow in Microsoft Word. We didn't have to get them to go somewhere else. And that was a huge learning for us is like, we wanted to build an electric bicycle for lawyers. They're already riding a bicycle. How do we just swap it for an electric bicycle where they don't need to change anything? And now they're just pedaling up these hills way faster. And that's how I felt about GitHub Copilot as an engineer. That combination of insights that we had from all of those years, plus that breakthrough created that amazing moment. And we were actually, yeah, it's As far as I know, we were the first company in the world to bring generative AI to lawyers in a commercial capacity before ChatGPT or really any other GenAI for lawyers.

That's incredible. And it's been working. You've gone on to raise $80 million total, which included a more recent $50 million Series B that was led by Keith Rabois at Khosla Ventures, which was at a $350 million valuation. So first of all, huge congratulations. Who was the first investor to say yes? Yes to investing in you?

I always get in trouble when I answer these questions. I answered one of these the other day and one of the investors pinged me. He's like, I think I said yes first. You know, it depends on the, on the stage. We had our first angel investor, Ethan, who's still on our board today, Ethan Buckman. You know, he just believed in this problem from day one. He had the same problems as an entrepreneur. He gave us a good early investment that helped us get started. We had a couple funds in our pre-seed, 3 Atlantic Canadian funds in our pre-seed, Concrete Ventures, Aquila Capital and Polaris. There's a very tight-knit community in Atlanta, Canada, so they were amazing. And then, yeah, our seed was led by, by Bling Capital, and that was actually one of our hardest rounds because we actually didn't have the breakthrough at that point. You folks know Ben and Bling, and they were our lead in that round. And that, in that round, we knew we were onto a problem and we were iterating really fast. We had an amazing team, but we hadn't really figured out PMF. We believe in the problem so much, Ben and the Bling team did too. Billy Draper was another person who believed in us very early on in that same round. And I remember meeting with the Bling team. They were super impressive. They had done some legal tech bets before. And they were like, I think you're going to be a cockroach. And I think you're eventually going to figure it out. That's one of their sort of investment criteria is like, will this founder just not give up? Or will this team just not give up? Up and they're like, we believe in the problem. We believe you're eventually going to figure it out, even though you haven't yet. That was a big inflection point for us. And they really helped instrument our company with the kind of quantitative rigor to keep measuring, do we have product-market fit or do we not? And then it became very obvious when we finally did. Yeah.

I love that level of conviction that they had, even when the product wasn't hitting at all cylinders yet, that they had the conviction in you. And of course, the vision of where it would go, which of course it has gone in that direction. And so today you have amazing customers. Definitely at the enterprise scale, from Nestlé to Crocs, Dropbox. But when you first dropped this newer product line, the waitlist exploded to over 30,000 lawyers in a matter of weeks. Bring us into those moments in the room around the growth inflection curve and like how you were actually starting to get this product in the hands of those lawyers.

Because we measured everything and we ran so many experiments, it was so obvious that something was different immediately. Even we would run little ad campaigns. We would measure like, what's your click-through rate? What's your cost per signup? And we just immediately see, it was not like, oh, let's wait and see, this is a hard decision. It was like immediately every metric in our entire funnel was better. And this really gave us in the company this culture of like using brute force to solve problems. Like one of our mottos is like, almost nothing works. That's our motto in growth is almost nothing works. We start everyone, every project we do is like a reminder, almost nothing works. Almost nothing really moves the needle. Almost nothing really matters. but our job is to find that 1 in 10 things or 1 in 100 thing that's really going to, going to matter. And then we started selling to these really fast-moving customers. So it was initially solo lawyers, small firms that were signing up because they could just get onboarded on the product very quickly. And then we got into some mid-sized firms that were also really fast. And selling to these customers who were like early adopter solo lawyers was amazing because they would just give us feedback so quickly. And we were optimizing at that phase for like feedback loops with our customers. So how do we get to customers like eBay and Dropbox? We actually never planned to sell to them, to be honest. We were going to sell to law firms and lawyers, and then these larger enterprises kept knocking on the door being like, hey, we really need a solution for this. I pitched Keith twice, and the first time I pitched him, he said no. It was one of the previous rounds, and he was like, I really think you should do in-house counsel, inter— inside enterprises. And I was like, ah, I think law firms are going fine. We're not going to do that right now. But eventually we let some of these enterprises on. We heard their feature requests, built everything that the small firms needed. We implemented a couple new features. Playbooks is one. So if you're in a— the big difference between a law firm and an enterprise is in the enterprise, you're reviewing like the same NDA like 1,000 times a year and you're executing the exact same rules over and over. Whereas if you're in a law firm, there's a much bigger kind of diversity of issues. So what the enterprises wanted was like this playbook where it's like every time we review an NDA, here's the 20 things that we do. So we built this playbook feature. We launched that last January and let on some of these enterprise in-house counsel and in-house legal customers. They just love that functionality. And then over the course of last year, in-house enterprise revenue has become like 55, 60% of our revenue now. So that segment has just grown massively.

Incredible. I mean, just to see the expansion of the core buyer through all of that iteration and learning is really interesting. And I'm assuming behind the scenes, there was a lot of iteration happening. On the product and ultimately the tech stack for how you were building the secret sauce of Spellbook. So curious where you've made trade-offs or decisions that have been strategic around leveraging other software providers and partners alongside what has been the magic, if you will, for Spellbook.

We are huge believers in leveraging foundation models, and we were very early on. One of the things I'm very contrary about, and I wrote a blog post about this bit in 2022, fine-tuning sucks, is super overrated and not needed for most vertical AI companies. Companies, and I think people got very deluded because a lot of VCs were like, oh, if you're just using foundation models, you're a wrapper, and if you're a wrapper, you're not defensible, these companies are just going to eat your lunch. And then, so you had all these founders hearing what the VCs are saying and selling back to them what they're asking for. So you had all these founders going like, okay, we're going to fine-tune models, we're going to custom train, we're going to do all this complex stuff. And we actually tried fine-tuning models as well. It actually sucks. It's not a very good approach in my opinion. Then we saw a lot of legal tech companies overfocus like this idea of building really complex, hard, defensible stuff at the expense of their actual customer experience. And a lot of these companies ended up already going out of business because they spent a long time training these custom models, and then GPT-5 comes along and it's better than their custom model. There's so many instances of that happening in, in many verticals. And so the most important decision for us in terms of tooling and what we're doing is like focus on harnessing the foundation models and then build around them You know, when you go from GPT-4 to GPT-5, you just automatically upgrade your entire product without any work. That's a real secret, I think, to how we've been able to do really well in the market. There's this incredible book, it's called Playing to Win. It's by this professional Street Fighter player, this like arcade game. And he talks about how like what the difference in the mentality is between the professional Street Fighter players that he played with and what he calls the scrubs or like the players who are not very good at the game. The Game. It's like such an amazing thing that you see everywhere once you read it, but he talks about how these beginners, their mindset is totally wrong in that they have this very romantic vision of the game that if they learn to play the game the proper way and the romantic way, they'll come out on top in the long run, whereas the pros in the game will use these really cheap tactics. They will ruthlessly hit the same button combination over and over again until they win, and it looks stupid and cheap, they're willing to take the cheap moves, even if it doesn't look sexy or if it's not intellectually stimulating. That's, you know, very relevant in what's happened in AI, where a lot of people assume that these wrapper products were just like doing something that was too cheap. And if you're not building your own models, you're going to somehow lose in the long run. And I think the exact opposite of that is what has, what has played out. That's the most important thing about how we think about tech.

We use like Braintrust for evals, lots of typical engineering of infrastructure in our product, but nothing super surprising beyond Thank you so much for giving us the rundown of your incredible founding story, the highs, the lows, the pivots, all of the good stuff that has brought forth an incredible company. And so with that, would love to hear what's next for Spellbook.

We're growing very quickly. We have over 4,000 customers in 80 countries, and that goes from small size firms, midsize law firms, large law firms, and also these large enterprises as well. Well, our mission is to get everyone in the world to better deals faster. And so that we can all just work together without spending so much time on the red tape and paperwork. And we think if we can do that, if we can create this 10x speedup, it will be incredible for the world's economy and all of our ability to just work, do the work that matters to us. Today, we accelerate maybe about a million contracts a quarter. We want to get to a billion contracts, um, probably at least a billion contracts a year, you know, is where we're thinking about how to get into and accelerating all of those agreements. I think we're only scratching the surface on the world's contract flow today. In terms of how we'll get there, I think one of the things I'm most excited about, we just launched a feature called Compare to Market. What this allows you to do is to actually take an agreement like a commercial lease in New York City, say Manhattan, and compare it to every other similar commercial lease in that same area using actual data statistics that you can inspect. And so we have a give-to-get model where our customers we can give these high-level statistics about what's your, like, late payment interest rate for commercial lease, or what's your— all of the kind of standard high-level terms. We have very deep statistics that we're able to develop from our data flow. I'm really excited about how transparent contracts are going to become. We are going to enter a world where if you receive a contract or I receive a contract, you're just going to be able to compare. I don't know, say you're like a hospital in the UK and you're acquiring software, you know, you're going to be able to be able to say like, is this a normal software agreement for the healthcare industry in the UK? Or is something really out of whack here? And you're going to be able to have the data to actually look at. And if something is off and you're getting screwed, you're going to be able to bring that data to the counterparty and be like, look, you're, you're asking me to sign a clause that 99% of other companies don't have. I can't sign this. And that, that's actually this data-driven negotiation, I think is going to become from just the normal way things are done. And this data has been so opaque and hidden, and negotiations have been so asymmetric for so long. Like, very large law firms have access to this data, large companies have access to this data, but most of us have no idea what the normal commercial lease in Manhattan is supposed to look like. And that data becoming transparent is going to become transformative because I think what we're looking for when we work with someone, if we were to start a business together or you were to buy Spellbook, is I'm not trying to get the best deal possible. What I want is like a normal deal. And what we want probably, if we want to work together on something, is just to shake hands and be like, let's do a normal deal. That's so much better than trying to butt heads and try to screw over the counterparty. No one's really trying to do that. Most of us just want to shake hands and then do a normal deal and move on to actually doing the work. I'm really excited about Spellbook being able to do that. I think we're one of the, we're probably the most used AI contract review tool in the world. And we have this incredible data flow that is going to really transform how deals are made and how people get to agreement. That's what I'm most excited about.

We're excited to sit on the sidelines and watch that happen. And just a huge congratulations to everything you've done and all the progress on sort of the front of AI and law and helping folks get fair deals. And with that, as we come up on time, I'd love to ask our hero question of the podcast. It's the question we ask all of our guests. Who's a woman in your life that has had a profound impact on you?

Pretty easy one for me. Michelle Sims. She worked with business incubator in Newfoundland where I grew up and went to university. And yeah, she oversaw one of the first companies I ever did an internship for, Avalon Microelectronics. That company came out of there and I got to know her there. Another company I worked for, Naukland, came out of there. And then my musical instrument company went through there and she mentored me through that. She eventually became CEO of this sort incubator, and then Spellbook ended up going through there as well. And the whole time, she just sort of mentored and, and was sort of like this wind at my back. And of all of these amazing companies that came from Eastern Atlantic Canada, she's definitely a hero of mine. You don't think there would be a lot of tech companies coming out of that kind of environment on an island on the East Coast of Canada, but it turns out there's actually an incredible entrepreneurial tech community there. She was such a huge part of that. She's someone I've worked with closely who's had a big impact on me.

Scott, thank you so much for joining us in the room today. It's been such a pleasure having you and such an inspiring story.

And thanks for having me. It was great to chat. Thank you so much for joining us at The Room Podcast.

If you want more from The Room every week, subscribe to our newsletter at theroompodcast.com/newsletter. We'll be back next week with a new episode. Inspirational guest Tuesday, 10:00 AM Eastern, 7:00 AM Pacific.

See you in the room.

Rippling is one of those tools I wish I had even earlier when I was building my startup, Pryve. If you're building a startup, every minute matters, but it's surprisingly easy to lose hours to onboarding paperwork, fixing payroll issues, or setting up laptops for new employees. We've ran into this constantly, and it's not hard work, but it's constant and it adds up. That's where Rippling comes in. Rippling is a unified platform that lets startups run HR, payroll, IT, and finance in one system from day one. Rippling's startup stack replaces disconnected tools that don't sync now with a fully connected platform. It lets you get back to the real reason you started your company. Right now, venture-backed startups can get 6 months of Rippling's startup stack for free. Head to rippling.com/theroom. Teroom and sign up today. That's R-I-P-P-L-I-N-G.com/Teroom to sign up for 6 months free today.

Perkins Coie supports the most innovative entrepreneurs and investors in fast-moving and high-growth sectors, addressing their myriad of legal needs. But the firm doesn't just provide end-to-end legal and business counseling to its startup clients. It also facilitates introductions to key advisors and sources of capital Perkins Coie's interactive website, Startup Percolator, offers access to programs, resources, and rich dynamic content designed to assist entrepreneurs on their startup journey. To learn more, go to startuppercolator.com and perkinscoie.com.