Mackenzie Jackson: Hello everyone. Welcome back to Cyber and Sake. I'm here with a really special guest I'm thrilled to have on, Maarten Mortier. Thanks for joining us here today.
Maarten Mortier: Glad to be here.
Mackenzie Jackson: So we're going to kind of get into a little bit of your history and kind of get into some topics. I'm sure we'll brush on AI at some point. But to start with, I lured you here with the promise of sake. So we're going to stop. We have two bottles here from different regions. So we have Sekura Masamune, which is a nice brand. This is done in a traditional brewing style. This is from the Kyoto region. And then we have another one which is a Ginjo style sake.
Maarten Mortier: The etiquette is that you hold it with two hands while somebody else pours. It's also part of it.
Mackenzie Jackson: I'm learning on this podcast. I don't know what the tradition is about drinking it, but I assume it's just chugging it.
Maarten Mortier: It's quite a nutty one.
Mackenzie Jackson: Yeah, let's get into it. I've got a bunch of questions that I want to ask you, but I think first for the audience that are probably unfamiliar with you, you were the CTO of Showpad, which is a very successful startup based here in Ghent. You've also been an advisor and kind of leading the product for another very successful startup, Henchman, here. And you've now entered into the world of venture capitalist, which is interesting. So where did you start on this journey? Were you software by background or?
Maarten Mortier: Yeah, actually when I go back to my story, I started programming when I was 11 years old, which is a pretty long time ago now. It's an embarrassingly long time ago. And I used to be always obsessed about creating. I was there when the internet started, right? And I loved to just make these things that were possible, that just started to become possible. I've always loved to make video games. I was never able to make a commercial video game. Maybe that's going to be something I would love to do one day if time allows. But yeah, that's my start. I've always been obsessed about programming, coding, and everything else is kind of still surrounding that part of what I love doing. I love the fact that with software you can just create and that there's no real big hurdles to do so. And that's still what I do, basically. I still I'm still very much involved on the technology side. Obviously there are chosen abstractions sometimes, keep it high-level to have other people do things, but I often code still, even when it's, or when I just think it's more effective. So that's still really my origin story is still part of my current story, even in VC. Definitely if you talk with me about your product, about your technology, you're going to go deep at some points. And that's still what I like doing and I still think software is pretty magical. We'll get into that later, but I think AI has made it even more magical. So yeah, that's my story.
Mackenzie Jackson: Well, it's a great story. I think it's a very perfect story to kind of end up in VC, because I think there's a lot of venture capitalists out there that kind of get this bad name for not understanding what they're investing in a lot of the time. And to have that technical background, I think it's an interesting combination.
Maarten Mortier: Yeah, and I respect people who go into it with a healthy naivety as well, right? When you don't know everything, you can still make pretty good decisions. Sometimes you can have less doubt or make better decisions because you don't know everything. So I respect all sorts of sides. It's like the same with a product manager. If they're not super technical, sometimes they make better calls, right?
Mackenzie Jackson: So you're based here in Ghent. I wanted to start with this kind of a random question, throw it out there. What's going on in Ghent? Because I came here for the first time when I joined Aikido. For those that don't know, Ghent is kind of a fairly small city in Belgium. It's a historic city. It's very beautiful, a lot of tourism. But it seems to kind of almost be shaping into the Silicon Valley, I mean, of Belgium, if not one of the major hubs of Europe at the moment. Why do you think that is? Why is Ghent such a spot at the moment for tech?
Maarten Mortier: Yeah, so the why? Because I really think the size of Ghent is actually pretty fundamental, because it's extremely small in a way, right? You can actually cross the city by foot in just half an hour or something. So, and I think I've lived and worked in San Francisco as well, and I think that there are analogies there where it's kind of like sizable. You can still grasp it compared to some other US cities, it's a lot more difficult to grasp just the sheer size. And I think as well, the size is kind of like just that there are extra dimensions of network effects happening where if you have the volume of Ghent, it just becomes like there's just a huge amount of network effects with that size. You have similar demographics, like similar age of people with the same problems and the same challenges starting businesses. And then you have a healthy layer of ex-founders, of what we call mafias of people who want to see this, who want to see this new potential, and they stimulate it with the right knowledge, which infrastructure is huge new building we have, the Winter Circus. It's just creating extra buzz, and this thing is in a non-linear phase, where it's just accelerating itself. And I do notice it. So, to be honest, I've always felt Ghent was very productive when it came to software. I know it's not very famous. It's still not very famous. But it's always been pretty productive, and I do notice a slight, yeah, non-linear, like I said, increase in that productivity. And I think we're just ahead of the curve where I think, generally speaking, people start businesses more often, more sooner. That's the thing you want to do when you're young. You still have that optionality within you, so you just do that, and that's the logical decision to make. And more people are taking that logical decision to start businesses with their friends or with people they trust. And that's happening at a slightly more accelerated rate, and I think Ghent is just at the forefront of a general global motion of starting more digital businesses.
Mackenzie Jackson: Yeah. I mean, and there's also a university here too, so there's also talent around, which is important for here is that, I mean, at Aikido, we need to grow the team very fast. We grew 120 people in one year, from about 20 people. And you can't do that if you don't have, you can get some outside talent, you can try, but I think having the talent pool close is…
Maarten Mortier: Yeah, and then you have a healthy kind of attitude towards university as well, because a lot of people when they exit university, they want a job. Some more cities who are built around their university have a lot of academic career paths. I think that's also a factor. Ghent does have academic career paths, obviously, but it doesn't exaggerate that. It's still a logical decision to really go into entrepreneurial private businesses and try to do something with your skills, try to grow them anyway.
Mackenzie Jackson: I wanted to twist this now into what you're doing at Entourage, because I mean, in some ways it's a somewhat classical story in terms of you were successful during various startups. You're then kind of able to take that knowledge and capital and reinvest it. But Entourage is a little bit unique in how you've set it up, in that you're kind of set up in two parts where you have this venture capitalist arm and then you also have an in-house studio to kind of build ideas or to kind of enrich your portfolio company. So how did Entourage start and why did you decide to go down this path?
Maarten Mortier: Well, we started out by thinking that that was a good idea, right? To just have that healthy combination. And I do still think every healthy VC should have some kind of incubator, some kind of building facility embedded inside their infrastructure. Because it just lends so much more, it's like what you said before, it lends so much more credibility to what you're doing as a VC. You're actually in the trenches sometimes, because there's just no proxy for when a founder has typical dilemmas that you face when you're building out digital businesses, you have decisions and there's no clear answer. And then I feel like just having a VC around that is really intimate with just the fact that you have these situations. That's already such a big perk to have that familiarity. And sometimes there's no clear advice that you give as a VC. Like, hey, you need to choose path A because that's going to be the best. Sometimes that happens and that's actually pretty cool to be able to do that. But often it's just showing that familiarity. So and you can only do that if you're credibly building your businesses. So there are two, I mean, there are two reasons why we do this, right? Because first of all, we think we can build these better businesses very well and have the right talent pool to this company. And actually build out a very profitable part of the VC branch. But there's also just the ability to show to the world that we are builders at heart, that we do not never want to stop building. And so that naturally arose when we were talking about how we can build a VC that is positioned towards others. That naturally became a thing for us, like, hey, we need to be building companies because it's just in our veins. So let's keep doing that. Let's not let go of that just because we're investing in international companies in general or pre-seed, seed stage companies. So yeah, it's a very natural decision for us. I do think it's sometimes it's more common than people think because a lot of VCs kind of abstract away that part of their activity in other business units, etcetera. So it's quite common to have VCs have some endeavor in the building out companies embryonic stage.
Mackenzie Jackson: Yeah, and I think even just talking to you for, you know, five minutes probably now, I mean, I can feel that it's kind of in your DNA from when you first talked about the origins of the internet. I don't think you're ever going to give up building something.
Maarten Mortier: And it's not just about the technology, just being able to box something to market it. It's so much fun to just have this new thing and have some excellence in that. I just love it. So…
Mackenzie Jackson: I want to ask a bunch of AI questions around this, but I'm going to wait. I'm going to hold off on the stick to the agenda. One of the things I'm interested about, and I promise to the audience this is going to come to some security questions now, but is around due diligence. So you're obviously investing in companies and also building them. So a little bit about my background, I was a co-founder and a CTO of a startup in Australia called Compargo. It still exists, but it's more or less a service company now. But I remember going through these due diligence stages with VCs and you put together these kind of, you know, these gigantic drives of everything that you could possibly think that they want to, but you have no idea if they look at anything, what they look at. So I'm taking this opportunity to ask you, what is it when someone first gives you that access to the due diligence folder, what do you go for straight away? What changes it? What gives you that gut feeling?
Maarten Mortier: Yeah, so gut feeling, and I don't want to sidestep the question. I will answer the question, but it always has to do with maybe the non-technical part has to do with the way the founders communicate and the interactions. So it's not just about the content of your data room, but just conversations with the founders, those are so much more important. It's something you cannot proxy with anything. It's just being you, being, telling your story. So that is really important. So that's where the gut feel usually comes from. Then some validation or some, yeah, just interesting interactions can occur in a diligence phase. Like the data room itself, how wordy it is, how verbose it is, definitely doesn't matter too much for me. I think in general we all know, we're familiar with as a founder, you're trying to get into this, it's kind of like a commercial deal, right? A VC's view is getting to the best terms, literally. So we know that you're trying your best to do that and you're giving data. To see that happening, to have some proactivity there, is already like, okay, interesting, but it's definitely not something that is super crucial. Now, we also have diligence phases where we ask the data in the shape we want and they give it to us. All of these are perfect scenarios to have pretty good information. The thing that gives me gut feeling, so maybe I could flip it around. What are some things that make me question the concept of the company is having some dissonance with the story? Like some integrity alerts, like when you hear something has been done in a certain way, has it, there's a certain story to a facet of the technology, something maybe is migrating, something has caused problems, and then that story is not consistent throughout. That is kind of something to zoom into because it's just very, and it is very sometimes difficult to verbalize what I mean, but it's just something you pick up on. Subtle signals where you know, like, okay, there's an interesting problem here. It's not necessarily something that will flip your idea, but it's just interesting. There is something here that has a story. There's a crack there where, and that's where the light comes in, like people say, that's where you see the founders also adjusting their company. They may have a story there that was interesting to understand how they react to some elements that are challenging when you're building a software company. In general, what I like to see is some scalability with the product. Like, does it scale? Is there a certain, is there a service component to it that was difficult to scale? Is there a good innovation story around using AI? Are they building something that has some moats towards the future? We might go into that later because all of these are very fundamental and very difficult things to just binary grasp. But you pick up on these things. But in general, maybe long story short is that I look at how we interact with the founders in that process. I don't really look at the data room asynchronously and green or red or orange lighted. It's just really, I prefer to have some extra questions, because that yields interaction and then we are like, okay, getting to know each other. Also from the other point of view because it's a commercial relationship.
Mackenzie Jackson: Yeah, there's two sides of it. It's so interesting you say this, because it's, I relate to it so much from my experience. And probably from a negative point of view in terms of we were trying to do a series A when we were going through it. We went through due diligence. We had secondary investors. We were in final negotiations with our primary investor. We're months down, due diligence passed. They kept asking for more documents. We kept sending them. And then finally, at the very last minute, we just got a no. And no one could tell us why, because I don't think they knew why other than it just didn't feel right. And I mean, like, we'll get to more questions about this, but I think that is an important part of VCs, and it might not be that the business is terrible, or that it just may not be that the fit isn't there.
Maarten Mortier: Yeah, sometimes you get outsourced due diligence processes and you feel that it's outsourced, right? When you feel like the VC used some other party. That can work, right? If they're credible corporations that they have done this before and there's a good, it can definitely, it's not always a bad thing. But sometimes you feel it as a founder and it's like suddenly there's this foreign entity getting through the standards or a questionnaire. And then you feel a lack of intimacy, lack of grounding with what you've been talking about with the VC. I try to avoid it. I try to keep the process because, like I said, it's a commercial process on two ends. Especially with good founders, good opportunities, you have to do that.
Mackenzie Jackson: Let's get a little bit into the weeds then. So let's say that you're interested in a company, you're going through the diligence process, great gut feeling, good problem, everything kind of feels right. When it comes to the technicality of the product, like the actual source code itself, are there things like from a security point of view or from a coding hygiene point of view? I'm not necessarily saying that you're going to review every line of code line by line. I'm more thinking kind of high-level. Are there things that are red flags or things that you really like to see in a startup's, or it might even be post-investment to say, "We really need to work on this product."
Maarten Mortier: That's good that you mention something like a post-investment stage where you already know about some parts of the roadmap and then you can, you can have an optimal cooperation later on when you've actually made the investment. So yeah, there are actually a lot of stages, right? Where we invest where you have pre-seed investment, seed stage investments. I think each of these have subtle different types of qualification, types of expectations, right? I think generally when it comes to, because you will split it up in code health, security. I think when it comes to code health, I try to focus on different things than what you typically learn in school, right? About good architecture, good software architecture. Because in the end, a lot of whatever is broken can be fixed. Especially I think we're in an era where refactors are becoming a little more easier to do than ever before. So, I like to focus on, is your tech stack equipped for hiring potential people to work in that? If you have a very exotic tech stack, you might have some, some difficulty there. And why is it exotic? Then I like to understand why, why is it because you have some personal romantic relationship with this type? Which can be a good thing, but on the other hand, and I know I'm going to speak on a lot of ambiguities, but around it can also be a sign that there's like a lack of thought. So, that is important. I also like to have tech that is LLM ready, because we're always going to mix the AI topics a bit, but some tech, especially if you have type-safe languages, and I know I think most software nowadays are in some pseudo or not type-safe language, you have a lot more benefit from using LLMs because they're able to use linters, they're able to use compile time activity to iterate on the output and able to protect junior engineers. So, I think there's definitely use, these two, this might seem like very out of ordinary focus points, but I think these two are actually fundamentally important at whatever stage. When it comes to health and typical architectural concerns like, do you do infrastructure as code or not? I can definitely accept it when you don't do infrastructure as code. I know it's definitely not a good position to be in wherever you really go at, for instance. But if you, if you know you're doing it, if you don't know what infrastructure as code is, I'm going to be surprised because that would be interesting to understand like, how are you deploying your, you have a CI/CD in this. Usually people know, right? And sometimes people have already quite an integrated roadmap for getting there, and then that's fine. That's equivalent to me to having infrastructure as code if you definitely know how to get there. So that's not really how I look at startups. I look at how found it be able to get there.
Mackenzie Jackson: I think especially in that early stage, right? I mean, it's not like if someone in there is like, oh, you don't have SAS implemented. Red cross. Like, no. You know, like all of these you're not doing infrastructure as code and Kubernetes clusters, right? And that's startups. Like, I know SendGrid for instance at Posts…
Maarten Mortier: Was it public or was it a quite acquisition? I forgot, but they had to rewrite the whole thing, right? They rewrote everything. And they did it post, so it's possible to be in that situation. So, I wouldn't say, I don't want to tell people to write bad software, not at all. But, it's just like where's your focus point? It's like the same with the test pyramids. Everybody remembers like if you do unit tests, integration tests, and end-to-end tests. I prefer personally, and it's opinion because you asked for opinions, I prefer people to spend a little more energy, much more energy with those end-to-end tests and like having some kind of safety net against really dramatic stuff and chaos tests etcetera, than spending this menial effort with trivial tautological tests.
Mackenzie Jackson: It's hilarious if you compare what you say, too. Is, um, I just, I only can really relate this back to my experience. But when we first built out our startup, we built it in Go, because we wanted it to be scalable to a massive audience.
Maarten Mortier: It's very difficult, so.
Mackenzie Jackson: It's very difficult. Also, we never got that massive audience that would justify that. Had we, and at that time, we're going back eight years, Go was brand new. Try finding a Go developer on a startup budget. You know, all these, they seem like, oh, I want to make these because it's going to make me investable. Like, I'm using the latest technology. I'm a hotshot. I'm fantastic. Actually, probably not the right, actually, definitely not the right decision for us. You know, and depending on someone else.
Maarten Mortier: If you're making web products, focus on what you're doing synchronously, asynchronously. You have to have message queues and consumers to really, and then you're also in terms of tech stack, that gives you a lot of freedom because a consumer, whatever technology you write it in, is not that important anymore. So, you can easily hybridize. You can have a consumer working in Go when it's super critical on IO basis or maybe on some kind of, yeah, error handling or whatever you have going on there. But, if you have a good architecture or what is being done synchronously, what is it being done asynchronously, how do I keep track, how do I have dead letter queues, and maybe some of these concepts are really pretty advanced, but that's fundamental. And I think sometimes when it comes to learning how to program, you should focus a lot more on those fundamental paradigms that are really happening a lot. Learning HTTP very well, learning how to optimize the use of HTTP, super important. But these are things you typically are able to, these are pretty simple things in a way. So, the bigger tech decisions are not like, should I use Go or not? It's like, what do I do synchronously, asynchronously? How do I set up? What's my CI/CD process? How can people experiment? These are really, really important things in a startup. I'm sure you guys were as well able to experiment easily in production using feature flags or whatever. These are like things that really push you forward as a startup, I think.
Mackenzie Jackson: Just to continue on a little bit on this track. When you're looking at investing into a company, you've mentioned the word moat a few times, kind of a barrier to entry that makes it difficult for someone else, kind of to come in. Slow down the competitors, let's say. I feel like the game's changing so much, so quickly. What are like, what do you see when you're looking at a company? What is a good moat to have? How do you and how do you, how do you judge that?
Maarten Mortier: Yeah, super interesting question and something that keeps me up in a good way, right? So, not necessarily I'm stressed out about it. I think I love the article that I think Google once wrote that it's a piece three years ago now and nobody has any moat, so to speak. Nobody has any IP anymore. Everyone has the same tools they use, they prompt engineer. So, a couple of things I want to pick up on that, maybe disparate things, is that there is still moat in having traction, right? If you cannot fake traction. Sometimes I think people sometimes lose track of the, there are a couple of really subtle things you can do with the software, the way you market it, the way you position it, the way everything kind of fits together. It's product related, it's also tone of voice related in your marketing. When everything comes together, and you see the initial signs, and as an investor I hope to see that very soon, maybe sooner than even the founders see it. That's rare, but I mean certainly as soon as possible. Then there's moat in just that strange confluence of being able to be at the right place at the time. And that's a very hazy answer. I know that's not answering the question, but there is still, there is still moats. There is still a capability to be on top of others just by having parameters come together. And sometimes you can pick them out. Sometimes you can say like, hey, they do this very well. Sometimes you can't really and it's just about, they're just having, just having a brand before you're even starting with having a brand. So I, and I, I do think we have to be, the VC, you have to be very keen on picking up on those things because that's where some companies that's maybe very deep due diligences would say like, hey, these guys have nothing. They're not doing anything innovative. But you can, you can see more than people see even though they're trying to look harder, you can see it better. So that's one thing. I think another, and that's more to your question. Another moat we should not lose is that we are currently like in 1996 when it comes to these LLM, usually a lot of LLM based pilots, right? It's…
Mackenzie Jackson: When was the crash? '99?
Maarten Mortier: Yeah. And I don't, I don't feel like it's heading that way at all. It's not heading that way at all to me. And even if it is heading that way, we still have a lot of post crashes, still have a lot of survivors, right? So Amazon survived through the crash. And their stock crashed. But 20x. So it went from $130 to $6. And they actually every day their stock decreased, they increased the amount of customers and amount of happy users. So 1996, what I mean with that is like LLMs are super slow. They are like insanely slow. I think, so they're slow tool calling with LLMs. I don't know if it's too technical for this conversation, but there's a lot of options you can do with LLMs to make them be more aware of specific, yeah, specific things they can do. They can go to the internet. They can go to specific ERP systems, do read and write things. That is still in its infancy. From a developer point of view, I think all of our LLMs as well are like, they're called LLMs, but to be honest, they're large number models, right? Because an LLM doesn't actually produce language. It produces vectors that have, that correspond to some token vocabulary. So, we will see more LLMs or more bigger neural networks being trained on other types of data, maybe AST. In your case, when you build, when you're scanning code bases, be ASTs are lower-level concerns being used to train those nets. So, they're still in their infancy compared to what I see, potentially in three years on, there's going to be so much faster, so much more accurate use of AI. You're going to be able to use AI fanning out to multiple AI models and kind of arbitrage models on top of that. So, in a way, our current AI native products, as an investor, it's actually an interesting time because in two or three years, they will be very much able to do magical things, I feel. Even though I'm quite an AI skeptic in a way, I do believe the technology has a lot of potential. So, they're still most because that's going to happen, and then there's really going to be a focus on how can I build a real AI native experience for my users? And I think we're not quite there yet because a lot of our experiences right now are kind of like, yeah, by consequence, they're trying to cater to what the users know in terms of software. They're using natural text, etcetera, to kind of make it feel like something that feels like AI, but still the experience itself is very much classical digital. And I think we'll have a lot more, yeah, more transparent, more like invisible things happening. Software that is working on your problem without you ever knowing that it ever did anything.
Mackenzie Jackson: I mean, I would like to go back to the like the Twitch had this gigantic breach with the source code where every single one of their repositories, 9,000 repositories, was leaked online. I downloaded it. Everyone could download it. I downloaded it because I wanted to see security stuff in there. The point of that is Kick already existed and some other kind of competitors. Now, there's not 100 Twitch competitors all of a sudden. And so their source code, their product, essentially their product wasn't, you know, that this, this, it's actually quite difficult even if someone gives you a product. So their IP, you're going to just assume that it's all out there.
Maarten Mortier: The brand is really strong, huh? By the way, people are sometimes dis-romanticizing the fact that the brand and you cannot engineer it. Sometimes something you have to be lucky with or something that, yeah, when they have the brand they talk about it endlessly on how they created, but in the end there are definitely things you can do that are, they could pick up and you could learn from. But it's also quite a magical thing to have. Suddenly it's there. Right time, right place, right coinage. It's just, it just works.
Mackenzie Jackson: So I think the kind of natural progression now is I guess to continue talking about AI because we need to. So I think perceptions have changed in AI when it comes to investing in companies. What do you think has kind of changed over the last three years? Let's say since OpenAI has kind of had its big splash and all of a sudden AI and LLMs went from conceptual to real. What like from an investor's point of view, like how, what has changed over the last three years when you're looking and evaluating at a company?
Maarten Mortier: Yeah. I cannot ignore the fact that building, so product development-wise, a lot has changed. And I would say for me, palpably, it's like the last year and a half. I think being able to use LLMs as a developer, also maybe as when designing and prototyping has really been more successful than I would have predicted. And it's actually changing, I think, the dynamics quite non-linearly. So I, I do think those effects are very important, being able to build product and development more efficiently than ever. Of course with the side effects. But in the end those side effects are sometimes exaggerated. I feel that there's definitely net benefits to using AI when you're coding. And the net benefit is also because sometimes you just have more positive energy from what you're creating as a developer. There used to be, I think, more stressful points in time when it came to executing a roadmap where there's, there's like ways to trade off that stress to hand it out and that's going to increase. You're going to have more like this feeling as a developer that you're actually controlling a team. So I think, I think that that cannot be ignored. That's a super impactful thing when you're building software. When it comes to the general ecosystem of building products, I think another thing that cannot be ignored is that the users are so much more eager to try something. People, generally in our business, because our world is fast-moving and we keep ourselves young with everything that's going on, are sometimes forgetting how the world was four or five years ago where using AI for your business was like, okay, I'm just going to wait it out until somebody else does this because I don't believe, I don't want to change. And I'm just using this ERP system. I know it's not perfect. I know people need 16 clicks to do a routine operation. I know I have work here, but every alternative is going to hurt me even more. That flipped in legal tech. That flipped in a lot of businesses that were like quite, it felt quite conservative. And what I love about that then is that it really creates a more product-led experience where you used to have startup scale ups really leading with good sales teams. Thought leadership and, yeah, trying to mentally wrap customers into a need. You have the opposite now where customers like, show me the product. I don't want to hear you sell one thing. I just don't, give me the sales pitch. Give me the, give me the, and I'm going to swipe my credit card. I'm going to, yeah, from one user going to 100 users. But show me the product. And that's what I love because as a builder, there's nothing better to have.
Mackenzie Jackson: I want to go slightly off script here and ask a question. This is one that's mostly from curiosity. And I want to understand how the landscape has changed. When AI first came out, I was like one of the ones that was most resistant to it, right? And I think part of that comes from like this fear of like, well, could this make me redundant? You know, like, could this replace me? You know, so your natural inclination is to be very skeptical, which I think is somewhat healthy. You know, you kind of need that. But when it comes to startups today, would you evaluate a company that was kind of not using AI in any way, not in their product, but potentially in kind of development, but just kind of like their product, what they were shipping, had no implantation for AI? Would that still spark interest in the VC world? I mean, you can't speak for the whole world, but you can't speak to yourself. How would that go?
Maarten Mortier: Yeah, I'm just going to speak for myself. I would be interested, right? So, when it comes to pure software, can you still make software that doesn't, that doesn't have like, that doesn't revolve around a why now what we call in using AI to solve a whole problem. I think that the answer is definitely yes. And it actually comes down as well to being in a new momentum phase where users are more keen on trying technologies and trying to maybe use new services. Yeah. You can actually sometimes build things that were able to, you were able to, sorry, build ten years ago, but you were not able to sell ten years ago. I do believe there, there's definitely something of that. There are also meta AI companies where you build software that services people who build AI companies, right? So, there's a lot of that. And then those, those don't always need to be deeply intertwined with AI itself. When it comes to not using AI to build companies, then I'm really more skeptical. I think if you, if you don't use AI natively to build out your company, I don't know why you're doing that. That is something I'm very skeptical about. Sometimes I hear about academic initiatives going private and then there, there still has and so on and I'm, I'm confused because if you're so hesitant on using AI in your coding, I've never heard a good reason to be hesitant.
Mackenzie Jackson: Yeah, I mean, and famously like Samsung and I believe Nvidia too, in the earliest stages completely banned, blocked all large language models like OpenAI and all of that. I'm sure they've all reversed positions now. It would be, it would just be prime for the turnover of your company.
Maarten Mortier: It's sometimes you have these things in VC, but anybody who is grounded into reality has this, some things are in French, they cannot be reversed. Some revolutions, there's not going to be, nobody's going to write CUDA without using an LLM anymore in future's time.
Mackenzie Jackson: Yeah. What is CUDA? Okay. I did see an article earlier that put the, one of the search terms that has increased the most over the last year is how to be Amish, and they've put that down to the fact that people are freaking out about AI.
Maarten Mortier: It's going to be some communities that won't use it, but it's…
Mackenzie Jackson: It's…
Maarten Mortier: Yeah, but in my opinion, the word, the world's AI has been chosen by whatever mimetic forces choose words. But the word is poorly chosen because we're not building, we're just building tools. We're not building, I think people are very concerned with intelligence, but it's actually completely ridiculous because they're just token generators. It just predicts, they just follow grammatical rules. And actually, nobody will ultimately understand all the decisions behind choosing this particular token, but it just happens to work. And I guess that's, we're just building a tool.
Mackenzie Jackson: The kind of last question I have, I want to take it kind of back to the VC funding, and I want to kind of cement it down. If you're looking at a company, let's say that they're pre-seed or seed, around about that. What are you interested in from a security point of view and post investment? Like, what, what would you like to see them, see them implement? You know, you've got a background as a CTO. So, what's the kind of path that ideally you think startups should be on at that point and be looking to?
Maarten Mortier: So, there are some basics that they, they need to be aware of and preferably already implement. Like being able to, yeah, have your supply chain in terms of code, like what you're, what you're bringing in from extra dependencies, validating, having some safety net there is really, I think, really important to have that because it's such a high ROI, such a low cost in a way and a high return to be able to have that. And it's actually quite risky, the supply chain, the typical supply chain models are actually quite, quite risky.
Mackenzie Jackson: Yeah, especially recent weeks has been…
Maarten Mortier: Yeah, and a lot of, a lot, I've been around for a while in software and actually the supply chain attacks will only increase from my perspective because a lot of them will be also using AI tools to have a much wider net. So that definitely will be important to have some kind of protection there. And also just being in general in control of your software bill of materials, like I say, being understanding what, what am I using? That is, that is kind of like a basic of a fundamental thing. I think in terms of your customer data as well, how you're using your customer data, you have to be aware of what you're doing. You have to be aware of what you're doing could be in violation with some of your potential customers. You might have an ambition to reach a certain size of customers, they have other demands because they are just structured differently and they want to have their own safety net so that that kind of trickles down to your own product roadmap and that's important to understand at least, even if you don't implement everything, and I'm talking pre-seed here, you're still building out something, you have to understand that, you have to be aware of what you're going through. I think in general, security is kind of a maturity process. I love actually looking at companies, I've done it before when, when we, when I was reading the acquisitions of companies as well, is that you look at the state of the art, the sort of the company as kind of like a maturity rating in, in different pillars. And the reason I like that is not because I'm like a McKinsey consultant trying to get this beta matrix out of it because that, that doesn't quite turn me on the matrix, but I like just being able to formulate the different steps and knowing what's the next stage is always good to have. There's always a next stage to be honest. And the one thing I don't want companies to do basically is having a security department that is going to deal with all of that.
Mackenzie Jackson: Yeah, I think healthy security is a bit like how good DevOps is, it just, it just has to be intertwined with everything you do.
Maarten Mortier: I completely, I completely agree with that, yeah. And what the managers, when they, when they start thinking about a feature, they have to, they have to stick about, is it's, this is always an unhealthy attitude to have some kind of facet of your business just wondering about these things.
Mackenzie Jackson: We have one more sake to try. I've had that one before actually. So I, I'd like to play a game with the guest here. The game's called, Would You Rather? I'm going to give you two options. I'm going to ask you a question and give you two options that you have to pick. Both options are going to be terrible, but you have, but you have to pick one. Would you rather never be able to use AI when building a product or technology or evaluating anything? So basically, we're going back five years. Or you can only use AI when you're coding, when you're writing anything. So you can prompt, you can get it to change things in the prompt, you know, you can get it to change the software, but you can't touch the physical code. What would you rather?
Maarten Mortier: I'm not going to cheat. I'm going to choose only using AI even though I have notes.
Mackenzie Jackson: Yeah, yeah, yeah, yeah, yeah. See, a lot of people try to manipulate this, too, and I'm like…
Maarten Mortier: I, to be honest, I would be there. But I have personal projects where I, where I like do crazy low-level as well. I just want to do it myself. But I want to keep that, that little part of myself. Professionally, I would definitely, yeah, choose the second because the second path is going to keep evolving as well.
Mackenzie Jackson: All right, we have another one. I'm going to kind of lean on your technical knowledge for this one, yeah? Would you rather have all of your secrets, your API keys, credentials, stored in your Git repositories? They are private, but they're stored there. Or have no firewalls on your application. So, everything in your application has to be perfect.
Maarten Mortier: Oh, man. No firewalls. No, no protection by cloud vendors against…
Mackenzie Jackson: No protection by cloud vendors.
Maarten Mortier: Then I would prefer the secrets being in there.
Mackenzie Jackson: This one, this one, this one often gets people. I get different answers from this. And I think people that have been breached by either option, it strongly influences their decision.
Maarten Mortier: Yeah.
Mackenzie Jackson: All right, last one. We've touched on dependencies already, so I'll do one based on dependencies. Would you rather have all your dependencies pinned to their current versions, or automatically upgrade to the latest always as soon as it comes out? So, if there's malware, breaking changes is implemented.
Maarten Mortier: Yeah, absolutely I would prefer pinning myself. I prefer pinning. Obviously, if something helps me have a more optimal process because it's kind of in the middle of your two options, then I would choose that. Potentially dealing with all the consequences that is not that whatever. Does sound stressful. I'm not going to lie.
Mackenzie Jackson: Well, thanks so much for coming on the show. It's been a blast. I really enjoyed you. I'd love to have you back. I have many, many more questions. I did have a feeling I wouldn't get through them, but, but yeah, and cheers.
Maarten Mortier: Cheers.
