The Business AI Was Supposed to Kill | Janus Klok Matthesen | Pixelz

Shamil Malachiyev: Good morning, everyone. On this week's episode of the Founder's Code podcast, I have Janus Klok Matthesen, co-founder and CTO at Pixelz, an international company founded in 2011 that helps companies retouch images and photos at scale. Hi, Janus.

Janus Klok Matthesen: Hello, hello, and thank you for having me.

Shamil Malachiyev: Before everyone sits down at home or goes on their way to the gym, can you tell us a bit about the current scale at which you operate at Pixelz?

Janus Klok Matthesen: Yeah, sure, I can do that. So like you said, we help companies retouch images and videos. And we do this for brands and studios all over the globe, mainly for e-commerce. We work with probably around 2,000 clients on a global scale. In this world it always depends a bit how much work we do every day - there's quite a lot of fluctuation in our workload - but every day we'll work with something between 30,000 and maybe 40,000 images, with peaks hitting maybe 60,000 a day. And most of what we do has to be retouched and returned to clients within -

Shamil Malachiyev: A day.

Janus Klok Matthesen: Well, 24 hours is the long one, actually. Most of our clients have what's called next-morning delivery. So if they send us images late in the afternoon, we'll have them back before they start working in the morning.

Shamil Malachiyev: How many employees do you currently have to manage?

Janus Klok Matthesen: We have, I would say, quite a small team outside of Asia. If I start with that: we're probably around 50 people in Europe and in America. Then in Asia we have two offices in Vietnam, one in Hanoi and one in Da Nang, and each of these locations has around 200 people. And then we also work with some partners in Cambodia and Bangladesh - that's probably another hundred people.

Shamil Malachiyev: I think we'll delve into a bit more detail about that in a little while, because for a lot of founders right now it would be important to understand how global outsourcing works. It's something a lot of people hear about, but not many people know the actual experience of going through it - what kind of decisions you have to make, what kind of complications you run into. In general, what do you think it takes for a company - taking your example - to survive for 14 years and to continue developing? I think that's the biggest question on everyone's mind, because everybody knows about the first two years, where you have to find product-market fit to start scaling. But what does it take to go that far?

Janus Klok Matthesen: First of all, the first few years - while you say we know about them, you are very right that those are super, super crucial years. Over the years I've been part of probably 10, maybe even 15 different startups, either just as one of the guys working there, or as a co-founder in several. Those first years are definitely tough, and it's also when you find out if you have product-market fit or not. And if you don't, you basically shouldn't try to survive - I would say you should probably pivot or do something different, because you need your product to fit with a part of the market.

After that, once you start achieving it, there's definitely a lot of grit. You need to just keep at it. There'll be so many hassles that you have to go through over the years. There'll be things that you plan for and expect to happen, and then there'll be so many things that you don't even imagine, that you have to figure out how to handle. So part of it is that you decide this is something you want to do and you want to make it work.

And then, for me, my biggest motivation ever, and what definitely drives me the most, is people. It's also, I have to say, what really keeps me up at night. So it has both sides. Surrounding yourself with people you want to be together with, you want to have these fights together with - I think that matters extremely, like crazy.

Shamil Malachiyev: You've mentioned product-market fit, and that if companies don't have it, it's not worth going all in. How do companies identify when they do have product-market fit? Everybody talks about this, but the exact moment when you say "we've got it" - how do companies identify that?

Janus Klok Matthesen: That's a good question. To a certain degree, I want to say there is definitely a moment where you start feeling: now I've got it. That being said, I also think you need to constantly be searching for it, to a certain degree. What happened to us was basically that we started the company very early on with the idea that it would be a small company we would run while we figured out what we wanted to do long term.

Shamil Malachiyev: Can you tell us about the small-company vision you had at the start?

Janus Klok Matthesen: I can - but it's also almost a little bit embarrassing. Basically we had three co-founders, and we were all in the previous company that one of the founders, Thomas, had founded. At some point he wanted to leave that company and do something new, and he wanted to do something global - that was one of his main motivations; the other company was quite regional and not easy to scale globally. Then me and Tais, the other two co-founders, reached out to him and said: hey, we love working with you, so we want to continue working with you. And he said: okay, let's do that, and let's just do something - because at that time he didn't have a job, so we needed something to pay him a salary. Tais and me continued in the old company and did this on weekends and so on. The idea was: how could we make a small business that could support us while we figured out what the big idea we were going to aim at was going to be?

And before we could even start finding product-market fit, we probably had to decide that this was the problem we thought we could solve. I don't think we really realized how big the problem was before we were lucky enough to have a super huge company reach out to us and start discussing how to solve their retouching pains together. Someone signed up - I remember it was from a Hotmail address or something like that, so they were not telling us which company they were from - but it turned out to be someone from Amazon who signed up for our service. It turned out they were looking for a service that could help them optimize images. They had started what was called Seller Central - I think back then, maybe it's still called that. Suddenly they were getting images and products from all over the world, from all these different retailers, and they were not controlling it themselves anymore. So they started seeing challenges with images, basically. We ended up spending more or less a year working with them and talking about how we could potentially help them solve everything. In the end, I think they had some cuts and needed to stop any project that wasn't really running yet, so we didn't end up deeply integrated into the platform. We still get clients from them, and I think they still like us - now it's more the specific sellers that we help.

Shamil Malachiyev: How big was the opportunity for you back then?

Janus Klok Matthesen: We don't even know - we didn't even calculate it. But I can tell you, we were in a meeting at that time - I think we were probably doing one or two thousand images a day - and Thomas, my co-founder, who was the CEO back then, was asked: what about scale? What scale can you get to? And boldly, he said: what scale do you need us to get to? And they said: how about a million images a day? And he said: if we can plan maybe half a year ahead, maybe we'll hit that. And then we didn't have that discussion anymore. I'm pretty sure we probably couldn't have hit a million images a day - that would probably have been impossible for us. But it definitely told us that there's a huge problem here that needs to be solved, that someone needs to solve.

And when you asked about the vision - I think that's really when the vision came to life. That's also when we decided: hey, maybe this wasn't that small project we wanted to do until we found a new big problem to solve. This turned into what we started focusing on for real.

Shamil Malachiyev: You've also said that you're constantly faced with challenges and you need grit to make sure you survive through them. And you mentioned there are challenges that are expected and unexpected. In reality, do those hit you differently - something you know is going to happen and you're trying to prepare for, versus something that comes unexpectedly?

Janus Klok Matthesen: I think so, to a certain degree, because the ones you expect, you can to a certain degree plan for. We all do budgets every year, and we try to figure out how the next year is going to look, so you have certain expectations. Some things you put in hoping they will not happen, and some of them happen anyway; some things you hope will happen, and maybe they don't. For me, at least, it's easier to prepare for those. And if you go out and listen to your podcast, or any other podcast about businesses, you hear certain stories that you can start to prepare yourself for. What I mean is that it hits a little bit less hard when you are expecting some of it.

And then you have the completely unexpected things. For me, there's sometimes a little period of shock when a big negative impact hits you. It can motivate you to really do something - but at least for a while, it can also hit you so hard that you don't even know what to do. A big one we all know about: COVID, and the more-or-less closing down of the world. That was definitely one we were not planning for, and we didn't have it in our contingency plans. Now we do - and I think more or less all companies have that somewhere now.

Shamil Malachiyev: Can you talk us through the COVID story? Everybody listening is going through similar things somewhere in their journey - it could be COVID, it could be AI, it could be anything. What was that experience like? What happened, what kind of challenge were you faced with, what lessons did you learn, what helped you persevere?

Janus Klok Matthesen: Yeah, definitely. Part of what happens with a crisis is that often you don't get the full overview, and COVID was a perfect example. We definitely knew something was going on. My whole team is in Vietnam, and in January - the year most of the world closed down, at least Denmark did - I traveled to Vietnam through China. In China I noticed they were checking my temperature. I mentioned this to my friends in Vietnam, and they said: yeah, we heard about this new flu in China, maybe that's what they're looking for. Then I came back to Denmark, and of course we heard something was happening.

We do these conferences - and the very first version of what we call our Flow conference was in New York. A big part of our American team and a big part of our European team were in New York for that conference. The conference was on Wednesday, and basically on Friday they closed down New York. So the first part was more or less just: how do we get people out of America now, back to Europe and back to their houses? In the very early days everyone was scrambling, just trying to figure out the problems that would hit us in the next two or three days.

I think part of what we do well in Pixelz, in a crisis like that or any other crisis, is that we're pretty good at getting together. We are quite a distributed team - I'm in an office where we are two Pixelz people, and outside of those two big offices in Vietnam, everyone is scattered around the globe. But we are really good at quickly getting online and meeting each other. That's what we did back then, a lot, to figure out what everyone was doing and what the next most urgent thing to fix was.

Shamil Malachiyev: Have you actively tried to facilitate a culture within the company - say, in the beginning stages, when it's mostly offline and you sit down with your co-founders and the initial team? What was the culture you tried to embody and pass down?

Janus Klok Matthesen: A big part of the culture we came with back then is still a big part of the culture in Pixelz, which I'm super, super happy about - because it's not easy. A few times I've scaled my team, maybe doubled or tripled it in a year, and it's super hard to keep your culture. And sometimes you shouldn't - at least you should be open to the fact that if you're bringing in new people, they're also going to bring something new to the table, and maybe what they're bringing is actually better than what you have. But it's a risk if you erase everything and start over, especially if you suddenly have more new people than old people who can carry the culture. So I'm super proud that parts of the culture we wanted to create from the very beginning are still part of our culture.

Relating that to times of crisis: one of the big things we always had as part of our culture is - I don't love the words, but you could almost say extreme openness. We're very, very open about what's going on in the company. To this day, we still share more or less all our numbers every month with everyone who's an employee of the company.

Shamil Malachiyev: What kind of pros and cons have you found in sharing all the numbers in the company? I've had a lot of founders and CEOs who are scared to share the numbers, because they're afraid that everybody is going to ask for raises, or ask what's going on. What do you think is the right situation in which a company should share the numbers and actually get everyone involved?

Janus Klok Matthesen: Of course, when I say we share all the numbers - we don't share everything; there's also private information in the numbers, even for individual people. But I do think, in the world we're living in today, when you decide to join a company, not many people do it just for the salary. Maybe that's something we can say at least in this part of the world, where most people don't go to bed hungry - but if we assume that's who we're hiring, and it definitely is for Pixelz, then people don't join Pixelz just for the money. They join because they believe in something we're trying to do, in their future possibilities and so on. And a way of proving that they can trust us on that journey is by being very, very open.

You're right that it also creates friction, and problems, when we have to share that we're not performing. Some people will get nervous when they hear we're not performing. Or maybe you don't agree with what someone is doing in another team, or their strategy - and that can put extra pressure on things, when you see the numbers and say: hey, why are you spending money on this thing when you should be spending it on that thing instead? So it opens up certain discussions, and you need to be ready to have them - including the discussion about whether people should have a salary increase or not. On the other hand, you'll probably get into a lot of those discussions anyway - everyone wants as much salary as they can get. In many ways it's probably easier for us, because most people in our company will look at us and say we're pretty trustworthy when we explain our numbers - because we share them also when it's not positive.

And that's what you have to keep in mind. We've had situations where everyone has cases of: do we really need to talk about this? And our CEO will say: hey, that's part of our culture. If we start not talking about this because it's a bit embarrassing - why did we make this mistake - it just doesn't fit with our culture, and it doesn't fit with the expectations our employees have of us. And if we really ask ourselves, it doesn't fit us either to put it under the carpet. Stuff like that has a tendency to start to smell over time. So it's better to get it out and just be open about it. That's my opinion.

Shamil Malachiyev: Have you noticed the way that you personally changed throughout this whole experience? Taking the Janus before Pixelz - what was different about you?

Janus Klok Matthesen: Good question. Together with Thomas, who I co-founded the company with, I've really been on a journey. I joined his first startup as his first employee. At that time, I was sure I was going to be an employee my whole life. That's where I came from, and that's my whole family - a lot of them are actually school teachers. I don't have anyone in my near family who started businesses. And then, together with Thomas and a lot of other people, I've been part of starting several businesses, trying to run them and scale them. That has definitely been a huge change in me. Today, when I look back, I can't imagine any other career where I would fit in. I want to be part of making the decisions too much, basically - and it's easier to do that when you own the company.

Another thing: before Pixelz, I didn't really think about myself as a leader. Through my school years and my studies, I wouldn't have called myself a leader. I've always been very social - I always had a big group of friends - but I wouldn't have looked at myself as one of the leaders. Now, when I look back, I can see that maybe my idea of what a leader is was different from reality. And maybe the whole concept of being a leader has changed over the years too.

Shamil Malachiyev: Can you talk about that a bit? Right now leadership is a big topic, and most of us just think: you take up your sword, let's go, let's conquer the new territories.

Janus Klok Matthesen: And that was probably also my mindset when I say I wasn't a leader - because I didn't like doing that. I'm quite a social guy. I love having people around me, I love doing things together with people. I don't find huge pleasure in doing stuff alone. When I look back at it - and what I think fits quite well with how modern leaders should think - it's being much more social, much more focused on: how can we unite a group of people towards a common goal? That's super important, and not so much who's in front. It's not important to me that I'm the one leading the battle. If I can see that I somehow helped unite these people so we can actually go into that battle - I love that much more.

Shamil Malachiyev: Can you elaborate on what you mean by being social? A lot of people just take it as: you talk to a lot of people, you have a lot of friends. But I think what you mean, in the context of the company, is not being the boss who hides in his office and does all the work - it's knowing the problems of the team, knowing people at different stages. Can you talk more about what you specifically mean?

Janus Klok Matthesen: It's true what you're saying. First of all, I find people interesting. I love hearing stories from people, and how they ended up wherever they are now in their life. I simply have a general interest in more or less everyone I meet, but definitely everyone I work with. And the same thing goes when I work with clients - I look at that very much as a social thing too. I really want to solve whatever problem they have, and I want to understand why they have that problem, why they built it like that. I think that gives me a deeper understanding of the real problem we need to solve.

And working with a group of people - in my world, I call them friends, because we end up being friends. Of course, I know it's a different friendship, because ultimately I could end up having to fire some of them, and ultimately some of them could end up saying: hey, I don't want to work with you anymore. But I go into it with the same motivation and thoughts I have in a friendship, and really try to understand the behind-the-scenes. That helps me when I need to relay a message, or explain what problem we need to solve, or set up a team of the right people.

I also love diversity. I don't seek people who are exactly like me - I actually work really, really well with people who are completely the opposite of me. It also goes for my group of friends: I love the groups where we are not all the same, where all these weird ideas come out because suddenly someone is thinking completely differently from the rest of the group. It just makes everything a lot more fun. Deep down, part of creating great teams is getting people to work together who are not at all the same - people who, at first view, you might think are going to have a huge conflict. If you get them to work together, and understand and respect each other for what their good sides are, and find a way where they don't end in conflict, you just get so much out of it.

Shamil Malachiyev: But it's so much work. You go in, and it's meeting after meeting, week after week - they start the conflict, and by the end of the meeting you've mediated it: there's our common vision. And then next meeting, the same. And you're like: oh, come on, just be friends already.

Janus Klok Matthesen: It's true - that can be tough. And actually, you asked what has changed in me: I think in the early days I would spend a lot of energy and time on those kinds of conflicts. Now I've become a little more direct about it.

Shamil Malachiyev: On that topic - a lot of founders I talk to, and I made this mistake myself 10 years ago: when you start a company, you love people and you love bringing people together, and you say, we're a family. And then after five years of people leaving, of having to fire somebody, the whole concept changes. Can you talk about your experience? A lot of first-time founders, especially in their first year, have that notion of "we're a family here". How can that be detrimental to actual business outcomes, and how did it change you as a person when you had people leave, or had to fire somebody?

Janus Klok Matthesen: Yeah, sure. Personally, I really don't like the family word. I also used it in the past, and I also thought exactly like that. But there is this one huge thing about family - I have children myself, and that's probably when you feel it the most: there is, one hundred percent, a very special bond to family. I'm not going to give up on my children no matter what. They're going to be there, and I'm going to be there for them - even if they kill someone, I'm still going to somehow try to help them through that. So of course there's a difference between family, friends, and business.

That being said - some of the employees in Pixelz made a Facebook group, which a lot of the former employees are also in, and it's called the Pixelz family. So I know we have it in Pixelz too. But whenever I get the chance, I always try to say: remember this. I like the concept of a family that comes together and so on - but of course we're not a family. We are people who professionally come together. That doesn't mean we can't build friendly relationships, and it doesn't mean we can't have relationships that are deeper than the professional level. But it means we have to have the professional level.

And coming back to those two who always ended up in a huge argument at the beginning of every meeting: at some point it's actually okay to go to them and say: hey guys, there's a lower bar here, where we need to be professional together. Maybe you two don't need to be best friends, but you need to respect each other's areas and communicate in a good way. That's okay to say.

I'm very happy to say that even the people I've had to fire over the years - most of them I still meet sometimes, and with a lot of them I still have really good relationships. That also comes back to the openness and being honest. Even when you have to let someone go, you need to be honest with them and tell them what's going on. In a sense, the easy one is when the company is not performing - sooner or later someone looks at you and says: now it's time, we need to cut down. It's not fun, but it's part of the job as a leader. The harder ones are when something is not working out between the company and an employee - that's a lot tougher, because then you have to get personal and tell people what's not working. But again: if you are open and honest, if you give people a real chance to decide whether they want to make improvements - if it's something where they have to change behavior, it's up to the employee to decide if they want to do that. And if not, it's almost like an agreement you make: okay, then let's part ways. That doesn't mean we couldn't go out and have a beer and discuss a football game - but it means we can't work together.

Shamil Malachiyev: I think the big takeaway for people leading teams right now is to think of a company as something where people unite professionally - it doesn't necessarily mean they have to be friends for life, no matter what. And you can still stay friends.

Janus Klok Matthesen: Yeah, I think you're right. I even had a situation in one of the previous companies where we hired probably my best friend from throughout the years - he's still my best friend. And I remember I was not hiring him; he was getting hired into another team, and I said very clearly: hey guys, I'm not going to be the manager of this guy. This is my best friend from childhood. And then, as things go, two or three years down the line some teams were merging, and suddenly I was the manager of this guy. It was super complex and super annoying. We ended up having probably a year, a year and a half, where we didn't talk so much in our spare time. I think he felt it was hard having me as a manager - and it was not something I had strived for either. Then he moved to a different team, I wasn't his manager anymore, and slowly we came back together. Today we are best friends again. So I've definitely seen how it can have an influence.

With Thomas and Tais, who started Pixelz with me - we've always had a pretty straightforward tone. In the early days we would have huge discussions, and people would almost be scared, because it got a little bit loud in an office or a meeting room. And then they would see us come out, go straight out and have a beer, and just be happy - happy that we could share our differences with each other. For me it's easy to switch between daily life and off time and the professional part.

Shamil Malachiyev: I think a very big secret that is rarely talked about is the ability to have a co-founding team that you can have a heated argument with, and make that a normal part of every day. If you don't agree, it's okay to have an argument - as long as you can come out of it thinking: I'm glad we had it, we managed to make a really good decision together, and the relationship stays good. It's the same as communication within married couples: you need to be able to handle conflict, because you're two different human beings who think differently. What would you advise founders who have co-founders about how to communicate when they're frustrated, to make sure they don't keep everything inside until it blows up one day?

Janus Klok Matthesen: Again, I would say the openness is super important. If you're seeing a pattern where something builds up and ends in a huge explosion, I would definitely bring that up and discuss: how can we open up for this a little earlier, so it doesn't end up being an explosion but comes out gradually? Also because you often need that information to make good decisions - not very late in the game; you need it as you move along. If it's something you need to change, you often need to push yourself into a new habit. A way of doing that could be: should we agree that we have a weekly meeting where we air these things for each other? Maybe you even force yourself to come up with three things you want to mention in that meeting. That could be a way of doing it.

And then, very much back to culture: if you actually get that into the culture of a company - I see it today. My whole engineering team is in Hanoi, in Vietnam. There's a big outsourcing scene there, a lot of companies doing outsourcing. And it's very clear to me that for a lot of these engineers, when they get hired into an outsourcing company, they are very far down the line of command. They get a specification that's a done deal - everything has already been discussed, and now it's just a matter of them putting in the hundred hours that were sold, hopefully not one hour more, because then the money goes out of the project. These engineers can be extremely good engineers, but extremely hard to work with, for me at least, and for our company. We have a hard time hiring these guys and getting them to work well in our company.

That's because we don't expect that everything we bring is perfect. We expect it to be full of mistakes - things we didn't think about. Sometimes things we should have thought about; sometimes super technical things we probably couldn't have thought about, but someone who's doing the work can question. We've built a culture where we expect to be questioned about these things. We expect to have discussions with our engineers: how is this going to solve that problem? "I know this client - it's not going to help them." Okay, then you need to hear about this other client, because it's going to help them. We spend quite a lot of time internally discussing every feature before we actually build it. If you come from a culture where you're expected to just do whatever the email says, just execute, it's a very hard culture to step into. On the other hand, we're also seeing people applying to work for us who want that mindset. In job interviews they say: hey, I heard that here you're actually expected to try to understand the real value we're providing for clients.

Shamil Malachiyev: And I think it's part of communicating that culture early on, during interviews. We have a very similar thing - whenever we hire engineers, sometimes it's an engineer from a huge company where they expect just a specification: they do the thing, that's it, no communication with anyone. But we insist: you have to be a communicator first, because it's a team game and you have to be involved. It's not just you sitting behind the screen.

Janus Klok Matthesen: Yeah - and this is also super interesting. I'm communicating this to my team these days: when we're having these discussions, when we're spending all this time, we're actually doing the thing that will have value in the future. When you're hitting the keyboard and typing in the code, I'm in doubt if that will have a huge value in the future. It's quite likely that an AI is going to be doing that for you. Maybe some of those discussions will even happen with an AI. But we need critical thinkers who can have that discussion, to come up with the right solution - because that's probably not going to be the first solution you come up with.

So I tell the guys: remember, when we invest time in this as a company, it's to build better products, to make something that's better for everyone. But you're also investing as a person. Whenever you do a job, you're investing your time, hopefully building something bigger than just the salary you get. And part of that is that when you take part in these discussions, you're investing in what a future engineer needs to be. I think this culture will also win in an AI era.

Shamil Malachiyev: You said earlier your main motivation in life is people. What exactly do you mean? Is it having nice people around? Is it having an opportunity to change people's lives?

Janus Klok Matthesen: It's probably a big combination, but definitely - I get energy from working with people. It gives me so much to think about all the good friends I have, and a lot of those come from the businesses I've been involved in. And of course, it's also a huge motivation to be part of changing - it sounds so big to say changing someone's life, but sometimes we are. There have been people in my career who changed my life with their input and their feedback. So obviously, sometimes we do change people's lives.

And it's wonderful to see - we have this Pixelz Pulse that goes out every quarter; at one point it was a physical magazine, it's not anymore. In it we have sections about coworkers who got married - where both people were working at Pixelz. Whoa, we have people getting married in Pixelz! And whenever people have children, there's a section about it, and it makes me super happy. The same goes for clients: when I can see the product really making a difference in their daily work. I actually count several of the clients I've had over the years in that friend bucket - people I still call now and then and hang out with. I just love it when you can make a positive impact on people - and likewise, they make a huge impact on me. That is a huge driver for me in life.

Shamil Malachiyev: With all the challenging times, decisions and pivots - what has been the impact of the last two or three years of AI development for you, as one of the leaders at Pixelz? What kind of strategic directions and decisions have you had to make?

Janus Klok Matthesen: From the product side - we started doing AI in Pixelz, creating our own AI models, seven years ago now. So we'd been doing it for a while before it was really big and everyone knew this was happening. It started mostly as internal work, to make all of our production a lot more efficient. Today, 60% of the manual processes are removed and are done by AI and automation.

Then the big moment was when ChatGPT hit us all, and we started seeing: hey, something real is coming out here. And in the last year or so, the image and video models have gone crazy. We were just discussing over lunch here that with the newest Nano Banana from last week, some of the images - you're not going to be able to tell they were created with AI. It's one hundred percent sure that in the future we can't tell, at least not every time. And of course that has an impact, no doubt about it.

So from the product side, we've been moving from internally focused projects to a whole AI team that is now extremely externally focused. You could also say we've moved from being focused on changing images to generating images and videos. That's very much the shift that has happened. We're sure that in the future we'll be offering our clients products where they don't send us the images - we generate the images, at least a big part of them. Probably we'll still receive images of the products, but then generate the videos of the model, or images with the model wearing the products - that could very well be done by AI.

I don't think it's going to be for everyone. I'm not talking 50 years into the future - I wouldn't want to do that, and probably not even five; that's hard at the moment. But for the foreseeable future, I think we'll see areas where it makes a lot of sense to use these products, and areas where it absolutely does not. We'll see brands who embrace this and really go deep - campaigns on the moon, or underwater, where no one is going to be in doubt that this is not something they went out and shot. And then you'll have others - I've already heard about the first few brands who have committed to not using AI at all. You're going to have those different sides at the same time, and we hope to cater to both of them. We think we can provide something great to both.

In the longer term - we just had another Flow event, in London. I wasn't there, but I got a few spoilers from some of the headlines. One of the speakers was from a Spanish brand called Mango, and they've been at the forefront of creating AI campaigns. It was super interesting to hear him. First of all, I think he was calling it augmented imagination - that's how he described the AI: it's helping our imagination come out; it's not artificial intelligence for his team. And another interesting thing: he was saying they're not doing it to save cost. Actually, those campaigns were more expensive to make than the ones where they shoot on models. They're doing it to create better content and more interesting content for their users. I think that will also be part of it. That being said, I've talked to a hundred clients who want to do it because they want to save costs - so that's also going to be part of it.

Our business is definitely going to change over the years. I still think creativity is super important. And in a way, this wasn't the plan: at some point we had huge challenges with production - at one point we were more than 900 people in the company, and just training everyone was hard. Back then everyone was doing full images from start to end, usually working for only one client. So we had a lot of small teams. Many years ago we changed this completely and built a completely new way of producing, where we split everything into smaller steps, and all our retouchers are now trained on these skills instead of single clients. When we started doing AI, it turned out this splitting made it possible for us to build AI that could work on just a small part of an image. We knew we couldn't make something that could work on everything - it would make mistakes. And in many ways this is still true today. The AIs make a lot fewer mistakes, but for most of our clients it's still not perfect. It will change stuff. And if you're selling something online and you suddenly start selling it with pictures that don't match what you're selling, that's a huge problem.

So somehow, in that process of redoing how we produce, we ended up building this hybrid system where we combine AI and humans. In the beginning, it was just to be more efficient. Now I'm more and more starting to think: hey, this is actually a core of our product - that we can combine these AIs with human skills, and thereby make something that's better than what you can do with AI alone.

Shamil Malachiyev: That's an interesting topic, because with the current view on AI - a lot of people I've talked to over the last two years, especially when new things come out, say: AI is going to change everything, we're not going to need humans, AI is going to be everywhere. They've created this huge bubble in terms of what the future is going to look like. Whereas in reality, there are a lot of hybrid models - and there are a lot of places where AI doesn't fit at all, where people are completely abandoning it because it doesn't provide any efficiencies. People say it's AI-empowered, but in reality you don't even need any AI for that. So if you look at the current state of the world - a lot of people see the new AI tools coming out and think AI will do the thinking and the making and everything. And when you look at real-world implementations, sometimes you have hybrid systems where it's 40, 50 percent augmented, and other times you find systems AI doesn't fit into at all. What's your take on the whole "we're in an AI bubble" question?

Janus Klok Matthesen: I kind of switch back and forth, and I think most people do. Sometimes you see something so amazing that you think: my God, they're right - we need this basic human salary, and then the AIs are going to work and we don't have to do anything anymore. And to some degree that's also a scary thought - I'm not sure how well adjusted humans would be to that. But let's see if that happens.

I get more and more convinced that AI is going to be part of all of our lives, just like mobile phones are today, and a lot of other tech - the internet and so on. And as that happens, there will be jobs that disappear, no doubt about it. But I'm also pretty sure a lot of this is going to go into us doing more.

My wife is a doctor, in - I'm not sure what it's called in English, but let's say the X-ray specialty. She sometimes tells me how, back in the days, shooting an X-ray took a lot of time - you needed to process it and put it up on that screen with the light in the back. You didn't do it just to check something quickly. And back in those days there was also talk about AI, because they started looking at this many, many years ago. She was told, I think 15 or 20 years ago, that there was probably not going to be a need for any X-ray doctors in five or six years. The reality is that at her hospital they're looking for more X-ray doctors. They need more. They have definitely gotten a lot better tools, including AI. It's a lot faster today - they can probably shoot 10 times as many images as they could 20 years ago. But the need has probably exploded by 20 times. Now, if your finger is hurting, they'll do an image - and she's going to hate me for this explanation, because some of it is unhealthy and they wouldn't do it unless it's warranted - but back in the days, no one would even think about doing any of those images for that. Today you do it, because it's so simple and so easy. And there's some stuff that helps the doctors describe it faster - today all her text messages, everything she does, she does by speaking to her phone, because that's what they do at work: they speak to these computers, they don't type anymore. So they got all those tools, they got a lot more efficient - but it also enabled the price of shooting that X-ray to go down. And suddenly, if your finger is hurting, why wouldn't we just take a picture and get it checked out? So I think we'll see a lot of that: because it's possible, the need will also grow, for many things.

So I'm not so scared about it. Back in the '90s, when I was studying computer science, a lot of the learning back then was about which language to learn - it was super important. You spent a lot of time storing all of that in your memory, because you needed to remember all the ways of writing things, when to use the parentheses and so on. Pretty soon after I was done studying, we stopped worrying about that, because we got IntelliSense and all of that, which would write it for us - if you could remember the first few characters and the structure, it would more or less fix it for you. And today it's moving even further: if you can just explain what you want to do and what programming language you want it in, it will do it for you. So that knowledge is not as needed anymore - but there are still so many things you learn in an engineering education that are super needed. My gut feeling is that it's going to be like that for many, many years still. But then again, there are also some super, super clever people who really think that in a couple of years AIs will be able to do everything. And I wonder - they're probably smarter than me - if they're thinking that, then how come... I guess we'll see.

Shamil Malachiyev: The thing is, from what I hear, when these people give interviews on podcasts, they talk about getting into those rooms with all the other highest-level people - and they're just usual people. Nobody can predict the future. I think that's what Warren Buffett says - he only has to make one good decision a year or so. Nobody knows exactly what's going to happen and how it's going to play out. Everybody's making bets - playing the casino, it seems. And as a CTO and an engineer at heart, what is your take on using AI coding assistants and how they're affecting engineering? How do you look at your engineers within the company - do you enforce it? Sonnet 4.5 came out, everybody get on it. What's your take?

Janus Klok Matthesen: I think if your listeners have listened to the whole interview so far, they will know that I wouldn't enforce something like that. I'm not very much for the AI manifestos - "if you don't do this, you're out and we can't use you". I don't think that's the way to play it. And some of the research coming out now is showing that at least some teams are actually shipping less code after implementing things like that.

But I did send out a message to all my developers about a year ago. And a month ago we had our developer summit, and a lot of my talk was about this specific topic. I'm more of a motivational guy, trying to get everyone to have a discussion with me about this - and in that, I also push them a bit, and make sure we have budget to buy some of these tools. Luckily, they are also super interested in this, just like I am. I'm not super happy with the vibe coding name, but I'm also doing some vibe coding in my spare time, making some fun small applications and keeping on top of it.

What we're seeing is that with the right use cases - and that also fits with not pressuring it onto people - engineers are just as motivated as everyone else to get rid of all the boring stuff we have to do every day. If we can get these tools to actually work for us and do a lot of that boring stuff, so we can focus on the interesting stuff - that's an easy sell, to be honest. I think the problem is that sometimes, when you let the AIs do the boring stuff, it ends up creating so much more boring stuff for you that you get demotivated. Part of my speech to them has been: I know this happens - I've seen it myself - and we need to not be caught in that trap of doing it just because it's AI, because now everyone has to do AI. But on the other hand, we also need the experience now, because it's improving every week, every month. There's a new version coming out and we can see those improvements. If we wait two years and then start using it, all the learnings we could have had in those two years - we lost those. So I definitely motivate very much, and we are using these tools quite a lot and seeing improvements from them. And we're also seeing areas where we say: let's not even try.

Shamil Malachiyev: Do you have any final advice for founders who are just starting, founders who are thinking of starting, or those looking at the whole future of AI and trying to find their own place in it?

Janus Klok Matthesen: That's a big one. Actually, my son - he's 16 years old - is more and more starting to look down the track where he might want to start his own business at some point. We haven't come to the point where he's asking me for advice yet, but I hope maybe someday that will come - or maybe he'll do his own learning. What I see from him, and from me, and from all the other incredible people I've been lucky enough to found companies with: being super curious, being open to investigating and trying out things, is just so important. So stay curious. Life is, to some extent, one long battle - so make sure you have fun battles, battles that are fun to be part of. You're also going to be part of some that are not, but then you can handle those. That's probably it. In hindsight, I'm so happy that I made startups and founding companies my career, because it's just super fun.

Shamil Malachiyev: Well, thank you so much for taking the time to share your story with us. A lot of founders listening to this now have a better perspective on how to handle their own interpersonal relationships within their companies, their own outlook on the AI world, and how to face the next day. Thank you so much.

Janus Klok Matthesen: I hope so. And I would also love to say: if you have any questions, if you want to have a talk, if you want my input on anything - reach out, find me on LinkedIn and just ping me. I love trying to help other founders. I love hearing stories from other founders. So reach out.

Shamil Malachiyev: Janus is going to do a podcast of his own. I will put all of your details in the description. It's going to be on YouTube, Apple Podcasts and Spotify, so anyone can reach out. We'll also do a big post on LinkedIn and a promo, so you'll probably see us there too. Thank you.

Janus Klok Matthesen: Fantastic. Thank you.