Matthew Scullion: AI is coming for £100m companies

0:00Welcome everyone to the Naked Founder On Location podcast, a business podcast where nothing is hidden. This episode of the Naked Founder podcast is sponsored by Financiale, the home of money for women. Today, the Naked Founder is on location and we're visiting the offices of one of the UK's active 41 unicorns. Unicorn obviously is a company with a valuation of a billion dollars and Matillion became one of those in 2021 when it raised$150 million in Series E funding. Matthew Scullion is one of Matillion's co-founders, and he's passionate about AI and data. Welcome, Matthew.

0:33Matthew Scullion:Thanks, Chris. Welcome to you. We're in my offices. No, absolutely. And we've got a fantastic view over Manchester in the background as well. God's own city, man. God's own. We always start every episode of the Naked Founder podcast with the same question, which is this. What didn't they tell you about founding a business that you wish you'd known? Well, probably how hard it is. yeah if you want to you know make money and get all the things that people perceive as being the bells and whistles of you know having a posh job then absolutely don't be a founder CEO it's very very long and hard and arduous and they're definitely easier ways to have a comfortable though.

1:14Matthew Scullion:And you've been doing it for a long time? Yeah, for this company, which has gone through multiple incarnations over this period of time, probably three major ones. But with Matillion, myself and my other co-founders, we've been involved with it for 15 years. So that's the overnight success, Chris. Absolutely. And then, yeah, I was also involved in another startup when I was much younger, ran that for a few years, it got bought. I worked for incarnations of the company that it got bought for again for a long time. So yeah, I'm lucky to have been involved in software entrepreneurialism, building businesses, leading teams for 30 years, Chris, which is a bit embarrassing.

1:59It's not embarrassing. You started incredibly young. You were 17. I'm going to go back to that actually. I'm going to ask you a slightly curveball question, which is this, is that This is Matillion, one of the UK's 41 active unicorns. Manchester's last unicorn. And yet when you walk in, you don't see lots of Matillion signs. What you see a lot of is Maya. Now, you're passionate about AI. In fact, your lanyard has got Maya on it as well. And you've invested hundreds of millions of dollars in Maya. And you talked about the number of reincarnations of Matillion. Is that one of your reincarnations?

2:36It is. Just explain what Maya is, what it does, and why you've invested so heavily in it.

2:41Matthew Scullion:Sure. So first of all, what Maya is. So Maya is really the product platform and the go-to-market brand of my company. The company is a corporate entity. It's still called Matillion. But what we sell, what we talk about, how we turn up is Maya. And what a customer would buy from us is Maya. And what it specifically is, it's an AI-powered enterprise software technology. Specifically, it's what we call an AI data automation platform. So what that means in English is you'll have heard of AI, Chris. Yeah, I think so. And you might even have heard the term, and if you haven't, it's an important one to get across, the term of an agent.

3:29Matthew Scullion:And an agent is AI technology wrapped up in a box to do something, to have a raison d 'etre and an end. It can solve a problem. It can do a task. That's an agent. Most all agents need or are going to need two bits of fuel that even the AI itself doesn't provide. They need data, information from a database probably or from some other source to feed into it. Let's say it was an agent that was going to decide whether you get a mortgage or not. Well, it needs data about you, right? It needs your credit score and your salary information and what are the bills that you've got. So that would be an example of data.

4:14Matthew Scullion:If it was an agent that was going to find a free slot in your diary and organise a meeting with someone. and it needs the other person's times that they'd like to meet and it needs your diary. So just two examples of data going into an agent. The other thing it needs is context. And by context, I mean if I ask my assistants and running mate, Laura, Laura's my EA, most people know Laura better than me, she's amazing. If they say, oh, I need to speak to Matthew about X, sometime fairly urgently. Well, she knows what X means. She knows who Matthew is. And she kind of understands what fairly urgently means.

5:00Matthew Scullion:Like, it doesn't mean in the next five minutes, and it doesn't mean in three months' time, right? She knows those things because she has context. And us humans, we do. We have context about things. And AI, unless you give it it, doesn't have context. So agents tend to require data and context in order to do their jobs right. The people in businesses that have been able to provide data and context, just like all of us humans, they're constrained in terms of how much work they can do. And we've worked in the data industry a long time, so we know that in general, the data teams inside of large companies, they were already struggling to provide all the data insights the data products, the data-powered use cases to their businesses before, even when that was just giving it to other humans.

5:53Matthew Scullion:But now we have thousands, tens of thousands of needs and opportunities to revolutionize the way we do business with AI and with agents, all of which need data and context. And you're going to end up in a world where you've got a machine-scale need for data and context, but only a human-scale ability to deliver it. So Maya, ironically enough, is an AI-powered solution to that problem. And what we mean by an AI data automation platform is a technology platform powered by AI that can provide that data and context to other AI agents and also to people and dashboards and all the things we used to use data for.

6:36Matthew Scullion:but do that in a way that is profoundly faster, more accurate, more productive, ultimately, I suppose, more cost-effective than humans ever could. And that levels up this machine-scale need to a machine-scale ability to fulfill the need. And that's Maya. And that's really, today, what our company is all about, providing that AI data automation platform to customers. Do we get to a point where we used to say, I'll Google that? Are we going to get to the point or are we at the point where we'll mire that? So I think we've been for a while now at the point where people say we will chat GPT or Claude or Gemini that rather than Google it.

7:20Matthew Scullion:So I can't remember the last time I Google searched for something, to be honest, because why would you just get it to provide the pages to read yourself when you can just have the AI contextually read it for you and give you the exact answer that you need, like a gazillion times faster? Or maybe a second-order version of that where it just does that before you've even asked. So if I'm going into a meeting, I will already have all the context that I need for that meeting pulled together for me by AI from various types of data sources, from recordings of previous meetings to what I can find out on the internet to what data we have on core systems here, so that I can be unbelievably prepared in the meeting, far greater than I could ever have been previously, because I would have never had time to do all the human work to do all that.

8:10Matthew Scullion:So I think we're already well past that. Everybody should be doing that. In terms of will someone Maya something, that's not quite what it's for. So Maya is a platform that would be used by companies to provide their agents with data. And so if you've got an agent doing something for you, and I gave you a couple of examples of what those could be from, you know, these are both fairly mundane ones, but scheduling a meeting or processing a mortgage request. but it could be like literally anything at all, then those agents need data to make that decision and Maya provides that sort of data. But you can absolutely use Maya technology to answer a business question.

8:58Matthew Scullion:So maybe you are the CEO of a company that makes outdoor apparel and equipment, right? And there's one of those in Manchester. and you want to know, oh, I'd love to better understand, you know, how the margin contribution of our breathable textile products is evolving over time and which channels are the most profitable for it and where should I take my product innovation next? Like you can literally ask that question to Maya and it would answer it with data and give you not just a graph or a chart without humans needing to build it, which is pretty staggering, but also then give you guidance on how to interpret that data.

9:44Matthew Scullion:So it wouldn't replace Google, but it would replace that sort of thing. And it's doing. I'm going to set the scene. It's fascinating because you've always been a really good teacher. I think the danger sometimes of technologists is they can talk above the person they're talking to. You never do. You keep it to a level that a simpleton like myself can understand. Just tell us, how much money have you invested as a business in Maya? Well, many, many tens of millions. And I mean, at the moment, I think we feel in a very exciting place as a company. If you were to walk around Silicon Valley at the moment, there's, I think it's fair to say, a bunch of people running companies of, you know, roughly the same order of magnitude as ours, maybe a little bit bigger or a little bit smaller, but, you know, proper technology companies.

10:45Matthew Scullion:And this is true even up to some very, very large ones, you know, the sort that own skyscrapers in the middle of famous cities. And the phrase that you hear quite a lot, Chris, is that they are looking for their second act. What I mean by that is that AI is, on the one hand, creating such incredible new opportunity for the way that people work and the way our economy operates, and therefore to build and create technology that makes use of it and participates in that future world that we're going to, but also so disruptive in changing the relevance of what we've all been doing for the last X years and you could argue X is like all the way back to the industrial revolution if you wanted to but certainly the last 10 or 20 and what people mean by looking for their second act is that they're like ah I was relevant in the old world maybe I'm not quite as relevant to the new world how do I reinvent myself for the AI native world that we're going forward in?

11:50Matthew Scullion:And they used this phrase, looking for their second act. So Matillion was, you know, our hope. One part lucky, two parts, lots of hard and smart work. But we started looking for that second act several years ago. And we were building our successor platform, which turned out to be Maya. we actually started that work certainly we were materially up and running at it in 2022 which is four years ago and we would talk about doing it in 2021 now that would have required some incredible forward-looking vision from myself and other people here in Matillion and one thing I have to say I don't like about that occasionally in my industry I think maybe all industries is we put down anything that goes well to our own individual brilliance don't we whereas there's a lot of lucky life.

12:42Matthew Scullion:And so part of the reason we were setting about that reinvention of ourselves was totally unrelated reasons. But thankfully for us, as we got started on it, contemporary AI started to come to the fore. And we could therefore say, well, we just happen to be getting started on rebuilding a next generation data platform. And this AI stuff looks exciting. Why don't we make our platform the world's first, best and only AI data automation platform? And we started doing that years ago. So everybody else is sort of, you know, not everybody, but many people and many famous and seminal companies are kind of thinking about this second act that they probably need to discover for themselves, invest in and make come true.

13:30Matthew Scullion:we're in that really strong and lucky situation where we actually were having that thought several years ago. And now as AI is really changing the world, we turned up in the market with the first best and only AI data automation platform, Maya. But yeah, it's many, many tens of millions. That's now paying a reward, is it? Very much so. And if we'd not done it, I'd be looking a lot more furrowed and a lot less excited. Yeah. You're looking great. And I'm feeling great. I mean, look, this market is crazy. And, you know, one of the things AI is doing is reducing the barrier to entry to develop software.

14:16Matthew Scullion:So, I mean, you and I were just talking about a version of this as we were getting ready for our podcast today. you were saying, oh, you know, maybe I should have a piece of technology that does X, Y, and Z. And I encouraged you to just go home and try out Claude Cowork and build it yourself. Now, if you can build yourself that piece of technology, there's a much lower barrier to entry to building technology. And macro that out to my end of the market where we're building commercial products, just the speed at which competition can move now is just so fast because you do it all agentically. I mean, Anthropic, who are the company that make Claude, one of the things they do which I think is just interesting and illustrative is they almost I'm smoothing out some details of the story here not least because I don't work there so I don't know it in detail but typically the way that we would have worked back in the day of making things out of metal is you know we'd have an idea we'd discuss the idea a lot then we'd draw it a lot of times to make sure what we were going to build is right and then we'd probably make a prototype and then we'd check that was right and then we'd draw it some more and then finally we'd sort of machine tool up and start making lots of it.

15:22Matthew Scullion:And actually in software it's pretty similar. We have a department in a software company called a product management department. Product management's job simplifying it is to listen and understand what customers need and translate that into product roadmap which we then build so when the customers get new features they solve the problems that they need. It's kind of like the drawing room of the draftsman of a software company. Anthropic now, they've totally reinvented that. They just build the product 10 times and throw away nine of them and keep the best one because the cost of building software has become so cheap because the AI is doing most of the work for you that it's more expensive to think about it properly and design it properly.

16:03Matthew Scullion:It's more expensive to do that than just to build it 10 times and pick the best. Do you think two questions? Because I want to talk to you about your founder journey. But the AI is such a... It's such an open conversation. But we're getting to the point now where we're getting multi-million pound turnover companies with one member of staff. Yep. And you think that will grow? I think yes, and established multi-ten or 100 million pound companies disappearing because AI makes their business model irrelevant. And zooming that out, in general, I think most of us, not all of course, but most of us are materially under-indexed on just how profound what is happening with AI is.

16:49I like examples. And we were talking off-air before the recording started, and you were talking about Mrs. Scullion. Yes. Secret behind my success. Absolutely, absolutely. And she said that you're a technologist, so anything with technology, ask Matthew.

17:04Matthew Scullion:Yeah. So she said, Matthew, I need to create a website for my business. Okay. and you spent about an hour and a half using the software that was available to create a fully functioning website with a picture of her relevant to her industry. Yeah, so, and lots of people are doing this, but yeah, first of all, I mean, just for our amusement, we'll reflect on the fact that anyone who works in technology or your relatives think you can fix printers and sort out the internet and, you know, why have I got all these viruses? You know, it's been a long time since I did any of that stuff chris if if ever in some cases and mrs scullion does actually uh know that uh in her defense but nonetheless um my wife's a physiotherapist and uh she spent a long time working in the nhs but she was um um you know sort of going it alone so she needed a website and uh and i think she was thinking of spending some money um on it i was like dude don't you don't need to spend money i was like just build it on lovable don't you and she's like what i was like yeah yeah let me show you this so we sit down and she's not got a logo so first of all we go on to appropriate AI tool and we just say need a logo here's what the business does here's the colors that she likes here's what the company is called and why bosh bosh bosh give me ten different logos you know 30 seconds later you've got 10 graphic designer quality logo she's like that one's perfect brilliant can I have all the different variations of that logo that could possibly need and a color palette to go with it please yeah done right so now you've got the logo and brand pack.

18:36Matthew Scullion:Once upon a time, even for a sole trader, that would have been minimum a few hundred, probably a few thousand quid to a local graphics design place, right? And then she's like, great, got a logo, got a colour scheme, need a website. So you go on to this site called Lovable, which everyone's using, so you'll have heard of. It's the fastest growing tech business in Europe, at least it was last year. Great business, lovely people as well. And, you know, Lovable is for building that sort of thing websites low fidelity prototypes it's like the ultimate vibe coding no technology expectation uh technology skill expectation at all oh it sues me chris so so yeah i'll go on to it i'm like right you know she's uh she's a physiotherapist she specializes in neuro rehab physio for stroke patients uh you know she does pilates part of the reason she's into those things she used to be a dancer and so she cares about body and posture and movement and muscles uh she operates from here she treats this sort of patient you know just put all this into a prompt here's the logo here's the colors bosh and you know a few minutes and a few tokens later out pops a brilliant website and then the other thing i was mentioning to you was um uh normally on one of those kind of sole trader style websites like the product's you isn't it so like if you had a website um it'd probably have a photo of you on it and you know we work in business don't we mate so you'd probably be looking like that.

19:59Matthew Scullion:But for a medical professional, you probably want to be in a medical setting, probably wearing the blue or the green or the white that medical professionals normally wear. And we just didn't have that kind of headshot of Caroline. So I go onto my phone, I pull out a nice family photo that we've taken recently, I pump it into an AI-powered website and says, can you just turn this into a headshot relevant for a website? She's a physiotherapist. So, you know, she needs to be wearing what physios wear in a medical setting, looking approachable but professional. 30 seconds later, out pops the professional headshot.

20:36Matthew Scullion:Now, none of what I'm telling you there is any news to you or anyone else listening. The interesting thing is the impact of that, right? Because she is a sole trader and she would have normally got out the yellow pages, I suppose, or got onto, you know, Google Local and she'd have found a photographer. She'd have found a graphics designer to do the logo, pick the colors. She'd have found some sort of marketing professional to help her with the copy and the story, and she'd have gone to a little website designer. You know, there was 10 of them in every town, wasn't there? And they would build her a website.

21:07Matthew Scullion:And we didn't need any of those four things. We did it at our kitchen table in two hours from end to end, and now she has a lovely website, right? And so it's macroing that impact up to the global economy that's the interesting thing. And that's where I was coming from with that story. And that's the point. It's like Block, Jack Dorsey just announced Block are going to go, we're going to lose nearly half their staff. Who's Block, sorry? Block, it's business in the US set up by Jack Dorsey, the founder of X, and he just announced that he's going to, like, half his staff because of AI. Yeah. And the point he was making is rather than have death by a thousand cuts, let's get rid of X amount of people on the right terms with the right way of doing it.

21:50He said, this is where AI is. Okay, I'm going to move away from AI a little bit, because we can spend the next three hours talking about AI. Okay, so there's a story that I love about you, and it was in the Sunday Times, 2021, when you were at Ultraman Grammar School, and you were going to take your French GCSE, and at the time you were described as a cheeky, slightly belligerent teenager, which I've never thought of you as belligerent, but your French teacher forecast you an E, and you went away and you spent 24 hours cramming, and you got an A, right? Is that story true of Matthew Scullion?

22:25Matthew Scullion:It is, yeah, yeah. I'm sure you've asked me if it's true before. Yeah, it's it. I don't think it's necessarily something to show off about, is it? Maybe questions the standards of French GCSEs. The one extra detail I will give you is that that teacher, he was a bloke called Mr Cleland. And I don't think he liked me very much in his French lessons. And looking back on it, with the benefit of adult maturity, I think he probably had a point. But in the oral part of the exam, you remember your French GCSEs? You had an oral listening, reading, writing. I was the only person in my year who was excused from taking French or any foreign language because I was so bad.

23:07Matthew Scullion:So this guy, he walked into the oral and it was in one of these little tiny rooms that you never otherwise went into in the school. And he had kind of a tape recorder and a little microphone, you know, a kind of proto-podcast set up, I suppose, Chris. And the way it worked is there were two potential questions he could have asked, and I'd crammed for one of them. So it was a coin toss, and it landed, luckily for me, on the right side, and I had a kind of pre-prepared speech in my head, and I think maybe it was 10 minutes. I could speak on this topic for 10 minutes nonstop. And about halfway through, he managed to find a gap and ask me a question.

23:48Matthew Scullion:And, of course, because my plan, Chris, was to one-shot it, that was an AI term, to one-shot it from end to end, when he stopped to ask me a question, I couldn't answer it. To be fair on this bloke, though, I don't know if he's still alive, I hope he is, to be fair on him, he could see that I'd tried and for the first time in two years done a stroke of work and he mimed what the question was. And you must be able to hear it on the tape, not that it will still exist, but I'm like, ah, oui, oui. And then I knew what he was talking about and I carried on with the one shot. And I ended up somehow fluking an A.

24:24Matthew Scullion:But yeah, totally true, mate. Well, when I was at school, I went to a grammar school in Kent as well and the French teacher got all the pupils that my classmates to write. Christopher has in Grand La Tête, which I said to my mate, I said, what does Tête mean? Head. So I said I've got a big head. I did have a big head. It was factually correct. But these days, that wouldn't be approved on. You'd be straight to the front page of the papers, wouldn't you? Absolutely. Okay, so this is the Naked Founder podcast, talking about founding businesses. Okay, so when you're 17, you did what all 17-year-olds do.

24:57You co-founded a software business. Okay, have you always been a person who doesn't see the risks? You just come up with an idea and you say, hey, tell you what, let's just found a business. You know, that's a certain type of person thinks like that.

25:11Matthew Scullion:Well, I don't know because I'm also in a totally unrelated data point apart from this one parallel I'll draw. I happen to not have brothers and sisters. And so people will often say, what is it like being an only child? And of course, the answer is, well, I have no idea because I haven't experienced the other side of the A-B test. And so I don't know what it is that makes someone more or less comfortable or less comfortable with risk or more or less likely to start a business or not start a business or do any of the other wonderful things that human beings do. In terms of risk, what I have said to a few people over the years is I do sometimes think that people don't necessarily quantify risk optimally for their situation.

26:07Matthew Scullion:And so with risk, one way of thinking about it is this probability and this impact. and probability is how likely is this thing to happen or go wrong or crystallise, whatever, and then impact is what happens if it does. So, you know, meteorites hitting the earth, very high impact, very low probability. Setting up a business and it going wrong, very high probability, but really, what's the impact, right? You know, you could also go to work and get fired, But no one does not go to work because they're worried that they might get fired. You go to work anyway. So if you set up a business and it goes wrong, well, it's kind of not that different from getting fired, is it?

26:56Matthew Scullion:You just have to go and get another job or start another business. So I've always wondered if people sort of maybe just miscalculate the risk. And they don't do it because they think there's a high probability of it going wrong. Correct, there is. It's a bit like relationships, isn't it? I'm not going to go out at that person because it probably won't work. Yeah, and you just end up being miserable. Yeah. Yeah, exactly, exactly. And then also, if you just try and think a little bit through a framework like that, then it can also reduce the probability of it going wrong. So we've kind of touched a couple of times on the fact that my business is now 15 years old, but it's not done exactly the same thing across those 15 years, and some cheeky journalistic cove such as you, Chris, might describe that as pivoting, whereas I just describe it as staying relevant and reinventing yourself and not denying what's going on in the outside world.

27:53Do you hate the word pivot?

27:54Matthew Scullion:No, not really. No, no, it's just... But it's just... I always think when it's a pivot, it makes it sound like we're geniuses, or whereas oftentimes you're just responding to the outside world, right? And then, like, the opposite, therefore, would be kind of irrelevance, wouldn't it? And those companies that are looking for their second act that we mentioned earlier, some of them might not find it. They'll become irrelevant and die because they didn't pivot, I suppose. Well, Blockbuster, Kodak, you know, the corporate graveyard is full of companies like that. Yeah.

28:32Here we are.