The new brand’s 12x summer #114

A warm welcome to a new episode of Noll till 100. We're back after the summer break. As usual, I have Fredrik Ankarsköld here on the sofa with me. How was your summer, Fredrik?

It was good, intense. In between, it was nice to be able to work on things that let you plan for new seasons, like the autumn, and catch up on things from before. When you start now, you don't have to run just to keep up the pace; you can start with a good push. Look at that.

Nice. How about you? Where do I start? First there was the union issue with Såpa. That ate up a bit of my… How did it go? We'll get to that soon. It ate up a bit of my holiday, but after that I was able to switch off. We travelled around half of South Africa, I was about to say, but southern Europe. We started in Croatia, then Italy. First my parents, then my girlfriend's parents. We ended up in Greece, in Crete. It was a nice summer, I have to say. Relaxing. Every entrepreneur knows the best ideas come when you're lying in a deckchair.

When you take it easy and finally relax away from the rat race, where you're running at a hundred miles an hour, you have time to reflect and think things through. With the silence, the calm and the stillness around you, a lot of good ideas come. You come back charged up, full of energy and with renewed enthusiasm. You have thousands of things to carry out this autumn and might carry out a handful. But that's enough.

Good enough. In a nutshell: a quiet, relaxing summer, and you're fired up. A new Monday, a fresh start. Let's go. Absolutely. I thought we could work through a bit of what's happened since last time.

A bit about our brands: what's happening, what have we done? Before the summer, you talked about having built a lot on an AI model, being able to prompt it much more specifically and having worked a great deal with AI. I think a whole episode on that would be extremely educational. But I want a quick update: where are we? We can get into Såpa, Mockberg and our new venture in the poster company that we talked about before the summer break. That we invested, or were going to invest, in a new company working with posters. How and why are we doing that? What's your thinking?

That sounds like a great idea. I want to start with the union. The ongoing union saga. Before we wrapped up the podcast for the summer, we were almost being hounded by IF Metall. It had escalated to IF Metall and Handels at national level to decide which union we should belong to. There was never any discussion on our side about not signing. The whole reason we hadn't entered into an agreement was simply that we'd grown very quickly and had thought about the business first, with very good, generous agreements for our employees. We hadn't been thinking that we had to join a union, but… We've actually covered that in earlier episodes.

Now we were basically getting constant calls from IF Metall. It came down to Handels responding: ‘We think you should belong to IF Metall.’ So we went back to IF Metall and said: ‘Okay, now we're ready to enter into an agreement. We just need to know how it affects the whole business model, our employees and their terms.’ We didn't want things to get worse for anyone, either the employees or the company.

IF Metall were very decent and said: ‘Let's talk about it.’ We went through what their agreement looks like, and they explained the different parts. We'll sign and finalise it straight after the summer, so everything takes effect on 1 September, if I've got my dates right. Somewhere around there, ish.

I remember that before the summer, it was more of a public pile-on against you. My God, it was, and it continued over the summer. But somewhere along the way it faded into the background a bit. People were still wondering, of course…

…but it calmed down. The campaign from the union's side was gone too. You've said you'll sign, so there's no reason. Exactly. We've committed, and the whole idea from the start was to do what's best for the company and our employees. It's never been anything else.

The fact that they even started a campaign against little Västerbottenssåpa was very strange to us. At the same time, you can see how the campaign against Tesla has gone. I think there were a few strikers left whom they bought out. Then there was nobody left on strike, and it was called off.

I didn't really understand that, that you could buy out strikers. We won't get into the strike there. If you're interested in Tesla and the strike, there are endless articles about it.

It should almost be illegal to pay strikers to stop striking. Buy them out: here's a year's salary, so you stop striking.

I know too little about it to… I haven't dug into it either. To be honest, as soon as we were done with the discussions, I thought: now we have to focus on the business. It's taken far too much time from the organisation, particularly the management team and the CEO. Now we're back on track: innovation, product innovation, creative quality. What launches are we going to do? Those kinds of conversations instead of discussing what's happening with IF Metall.

What have you come up with? What creative things are you working on? Product launches. We have a lot to talk about, but I don't fucking know how much I can say on the podcast at this stage. What we can say is that we launched an anniversary box over the summer. It was the 100th anniversary of Västerbottenssåpa's recipe. It's exactly 100 years since the recipe first came to life. We thought that had to be celebrated somehow, so we developed our own anniversary box with a hand-painted cloth, completely new packaging and new, very festive labels.

It was produced in a very limited edition. Not many people could get hold of it. We saw enormous demand and sold out even before the product was ready in stock, which was pretty crazy. Very high demand, in other words.

But the delivery went well anyway? It started about a week after we published it as a SKU on the site. It went well until we saw that a SKU was missing and needed to restock. That's always how it goes with logistics and products. You do the best you can. We sorted it out in-house, and eventually all the customers who had ordered their anniversary box were able to receive it. As far as I know, the vast majority were exceptionally happy and thought it was a good box.

Now we're brainstorming something for Christmas. We'll probably sell something for Advent or Christmas, a similar box. I was sitting there thinking: how can you have an Advent calendar for…

…Såpa? No, you can't. Last year, we had an Advent calendar where you got a cleaning task every day until Christmas, so the house would be clean and tidy for Christmas.

It was quite a fun thing. You got tips and tricks: now we'll do the fridge. Think about these things to get rid of smells, clear it out, switch it off and start it up again, and so on. Now we'll do the oven. The next day we'll do the shower. Every day we had a challenge.

People who followed the whole Christmas calendar, which they also received in a physical format, got a little gift from us at the end. How is Mockberg doing, then?

It's going incredibly well. Mockberg is doing very nicely right now. We're just about to open, or have just opened, our shop in Berlin. Our first flagship store outside Sweden.

Congratulations! Thank you. Of course, that's not down to me. Who's been responsible for it? It's a shared responsibility. We have Elvira, our CEO and founder, and Daniel, who's come on board. Her brother was involved before, took a break and has since helped us get the company organised…

…over this latest period. And of course the rest of the team too. We have an amazing group of women at Mockberg running the company, and a good team from PG led by Domantas, driving the growth work. The organisation is growing, and we've hired more people. It'll be fun to see how we roll out the shop and how it's received in Germany. Mockberg doesn't have the same brand awareness there that we had when we opened at NK in Sweden just over two years ago.

That shop has really taken shape and improved over time. We hope and expect the shop in Berlin to be much the same. Growth outside Sweden is very good. In Germany, we're growing by triple digits compared with the previous year. That's very nice.

Why did you choose Berlin for a shop? Germany is a large market outside Sweden, and it's done very well for us online. We can see that our online penetration can help the shop. We've created a community and have a strong base in Germany who already know Mockberg. Historically, Mockberg was rolled out through retailers, before my time.

Before I was a part-owner, it was rolled out through retailers. Are they still there? They're no longer there. That was largely pulled back during the pandemic. Since then, we've tried to bring the retailer channel back too. We have retailers that are doing well in Germany, overall. But we'll see. It's an incredibly exciting time for Mockberg, with lots of launches and lovely products that came out over the summer, and new initiatives coming now. We'll see how they're received. A big thing for me is how much fun it is to go to parties and see… I was at a joint 40th birthday party for a good friend and his wife. When you look at the women…

…half the women at the party are wearing a Mockberg watch. It's a milestone when so many people around you use the products. I think that's amazing.

An amazing feeling. Absolutely. Of course, that's not my doing; it's the team, led by Elvira, who have developed fantastic products…

…that basically sell themselves. No, of course they don't, but you can almost say that. It starts with the product. If it hadn't been attractive and good, we wouldn't have been able to scale the brand. You mentioned the deckchair. Is that where the best ideas came? What have you been pondering there?

A lot of different little things, but the big underlying question is: how can we apply AI and get better at it, both in the business at Porobic Group and across all the brands? How do we get better at AI on the creative side of Mockberg? How do we get better at Såpa at creating agents, better dashboards and reporting? There's so much you can do. How do we do it in practice? We've tested and played around a bit, but you need a clear strategy in your head. Before you start prompting and building, you need to know: what's the end goal? What do you want to build? A lot of it might not exist today. How do we build something that doesn't exist when we don't have inspiration to draw from? That's…

…where my thoughts have been. You've built an app, actually a whole system for e-commerce businesses. We talked a little about it before the summer and haven't touched on it since. How has it gone? I know you've tried to connect a lot of systems. For me, one of the things is what's called a ‘harness’ in English. I don't know the Swedish word. I read about a test where they took an AI that was in the top 30.

It wasn't capable of all that much, but if you put it in the right context, it was suddenly in the top five. How smart the model itself is isn't the most decisive thing. Of course, that makes an enormous difference too, but the tools it has access to, the things around it, make a huge difference.

Are you with me? No, not really. You have a dumb model that becomes smart. Explain how. Let's take an example. You say: ‘I'm going to build something with Claude for my company.’

Then you have its software, or Codex. We can compare it with Lovable, which is a wrapper, sort of.

But it's more than a wrapper. It feels like a wrapper, as does the software I build. There are two things I'm making: that piece of software, and another to help me stay focused.

For some mysterious reason, Claude isn't the best. If I ask Claude to design a page without clear instructions, I get a mess back. By a mess, I mean I look at the page and don't understand anything. I can't take in the information.

So it's nonspecific and not relevant to you? No, but then I discuss it. I've researched it: how come I ask Claude or Codex to build something I then can't use? It's overwhelming; I can't take in the information. In this case, it doesn't have the design context. Even if it knows design, it doesn't have the framework. If you give it no rules, it designs a box inside a box inside a box.

If I say ‘Design me a dashboard’, I look at the dashboard and have no idea what I'm supposed to do with the information. It doesn't give any concrete actions. I even asked it why. It was a long discussion, not five minutes, where I tried to understand: why can you build a dashboard I don't understand, while my own software builds one I do understand? The explanation was that Claude, if you give it no instructions, designs like an agent, not visually. It designs from the agent's perspective. For an agent, a box inside a box inside a box, where everything is…

…grey, is a hierarchy in the code. In the code, the dashboard is very logical to the agent. But to a human, everything looks exactly the same, just grey boxes inside each other. So you don't get a clear hierarchy of which information matters.

Or it builds several buttons in strong colours, so you don't know what the primary action on the page is. What I mean is that if you have what you might call an AI wrapper, an LLM wrapper… I also build Claude into my software, but I've trained it in design.

Things like design hierarchy, and given it loads of information: this is a good page, this is a bad one, this is a good one. Exactly. I think I burned over a billion tokens just getting into design. A billion.

That has led to, today… I'm building another piece of software to help me stay focused. Like [unclear in the Swedish source: likomo], which I've mentioned before on the podcast, it can limit time during time blocks. But what I'm building into my own software, a desktop app, is that it can also see what I'm doing in Chrome. Not just block specific sites. YouTube can be blocked, but if I'm using it for a tutorial related to an action I'm going to take, YouTube is a good thing. That's why I'm building the external software.

When I wanted to build it with both Claude and Codex—oh my God, the amount of time I spent. They made boxes inside boxes, buttons on buttons, and crammed a crazy amount of text into the dashboards. I was like: what does this have to do with anything? That's what I mean by a harness in my case. I'm no super coder, so I might be expressing it incorrectly. But based on the way I work, I've given it a crazy number of design rules, things it should do and examples of what's good and what isn't. When it builds through my own software, it knows: I mustn't build boxes inside boxes.

I'm not allowed to write out text explanations. Another thing AI does is that it loves to describe something by saying what it isn't. Sometimes, if you ask Claude to write a text…

…it often writes: ‘You don't do this.’ You're like: but what do I do? So it has rules such as not making boxes and not describing things through what they aren't. If you describe something, you have to describe what it is. I also have lots of design rules. One is: if you make a design and feel—or not feel, because this is data, but interpret from the analytical data—that something needs explaining, then the design is bad.

If you make a box that needs masses of explanatory text around it, it's not intuitive. Then you need to improve the design until it doesn't need any explanation. Those are the sorts of things I mean that can sound like an LLM wrapper.

That's the work with a wrapper: taking a big, general program, a general AI like Claude, and narrowing it down to do only this…

…and then training it on that and giving it loads of… I don't think ‘wrapper’ means anything negative. That's what I mean. A lot of people might think: ‘Why should I pay extra for that?’ Of course you pay extra for it, I think.

It's so funny, because I was talking to another guy who uses similar software. He said that when he uses the software compared with Claude, he reaches a finished result faster, with fewer iterations. It might be more expensive in itself, but if you need fewer iterations to get the result, it becomes cheaper in the end.

It pays for itself. Exactly. That seems reasonable. A wrapper really means you pay for someone else having trained it and specialised it for your particular purpose.

Often, you might get a better UI too. That's partly what you pay for with Lovable: a good-looking UI, the fact that it saves the sites and has connected things after the site is finished.

You might need somewhere to host the site, or something else around it. You have the ecosystem for building the app or website. It's not just building in Claude, where it gets far too general or something else.

It's the same when I build with my software. If the user is from the US, it needs to be stored in the US. The data, or…?

The data, yes. Compared with if it's from the EU, then preferably it should be stored in the EU. That's a separation too. That's the difference: if you build it yourself, it can easily become a mishmash. I see.

The issue I see is that AI gets better over time. But your software will always be AI-agnostic, I assume? You won't be tied to only using Claude. No, you can switch…

…between them. If ChatGPT, or OpenAI, has the best model right now, OpenAI can provide the logic while your wrapper still looks the same. Then OpenClaw comes out with a better model, so you can have the latest model. Then Anthropic comes out with the latest one, and you use that, and so on.

That's an advantage of a wrapper, compared with being committed to Claude or something else. That's what I think is so amazing today: being able to use wrappers and not see them as something negative, as with Lovable. A lot of people don't think about the fact that when they're building an app, Lovable has also sketched out an enormous amount more UI and UX data than the agents. I don't know why that is, because in my world that's what they want you to build…

…but it still isn't built in. I don't understand that logic. One thing I've been looking for in AI, which I was thinking about in the deckchair, is a project where I buy a Mac Mini.

Then I download something like OpenClaw or some kind of database to create my own agent. The agent should be able to connect to the computer and always have access to it, including different applications, whether it's email, your whole life or social feeds. So you can chat…

…with the assistant, which adjusts things in real time. You can change a meeting or have a conversation: ‘An email has just come in from someone who'd like a time on Thursday afternoon, but I can see you have something then. Would you like me to move it for you?’ ‘Yes, you can do that.’ Then it emails everyone without you having to do anything. You have a chat window or even a dialogue, almost like a conversation.

You can build that yourself. It's a bit of a project. I tried it in May after talking to Jakob and hearing about his ambition to build it, so I built one too. There's a clear tutorial on YouTube. It took about a weekend…

…to sort out. I'd say I'm fairly technical and knowledgeable. Not the best, but six out of ten. Think of everyone who's a bit less capable, or doesn't have a weekend to spend. A Mac Mini can be fairly expensive. Why isn't there a computer where you buy a ready-made agent? You have 100, 200 or 300 integrations ready. You don't get the bugs along the way and don't have to set anything up.

The two days it takes to set up have already been done, ready out of the box. You might pay a one-off cost of 4,000, 5,000 or 6,000 kronor. I'm guessing you don't need the capacity of a big, hefty Mac Mini to run the agent as it is today. It's future-proofed so you can update to the latest AI model. Right now it's OpenClaw, but tomorrow it's something else.

There are some under Hermes or something similar too. Right. I've been looking for that but haven't found anything. If any of you listening have a tip, give me a shout and I'll take a look.

Have a look at Hermes. That's one of them. But is it a ready-made box you can buy, or software you download, like OpenClaw, and then do the work yourself? Yes.

Then you have to instruct the AI and connect it up yourself, at least a little. You have to fiddle and tweak. It becomes a weekend project, in other words.

I want to warn people about what happened to me this summer when I was building…

One of the reasons I'm building the focus software was that I got what I think is called ‘AI fry’. It's similar to the term ‘brain rot’ that you can get from spending too much time on TikTok.

Your brain starts rotting. You get too much dopamine and downregulate the receptors, so you need more and more to feel anything. Eventually, some teenagers are almost lying in bed and can barely do anything. Then the solution isn't more, but rather a detox so you upregulate the receptors again. What I got into was ‘AI fry’, I think it's called.

I got like that from working with AI and started researching: this is starting to make me feel unwell. There was one thing I hadn't reflected on. That's why I want to warn others so they don't end up there. I could ask Claude: ‘I'd like to change the colour of this button.’ Later, I took all my chats and imported them into another AI to analyse, because I felt I was getting less and less done. Eventually, I didn't even want to sit with that shit…

…any more. The explanation was that what it was doing, which I hadn't reflected on, was: I could say I needed to change the colour of a button. While it was doing that, it found things in the code. First of all, it dumped three A4 pages of information on me without my asking. It didn't show which information was for me and what was internal dialogue. I was being exposed to extreme amounts of information, and it became overload. When you're sitting there all day and don't know: what am I supposed to do with this information? Which part matters to me? Is there an action I need to take?

Am I waiting for something? Another thing was that I'd think: today I'm going to work on this task. At the end of the day, I felt completely drained but hadn't done the task. I couldn't put my finger on why. When I traced the history, I saw that Claude itself had sent me down 17 side tracks. I asked for the colour change, and then it went through the code and found: ‘I found a bug. Do you want me to fix that too?’ Yes. So you start getting stuck on bug issues.

Eventually, I built a rule into Claude: if there's a bug, don't even ask whether you should fix it. You don't need to ask whether you should fix the error. I've asked it not even to ask me about lots of those side issues.

You know what I think, Fredrik? You've become an AI power user. You have a luxury problem that not many people can relate to. I use a lot of AI…

…but I absolutely can't relate to what you're describing. You're also building software with AI that's meant to help other people. I don't think people can relate, but it's really interesting that you can end up on that sort of side track. AI is basically building your day.

Exactly. AI controls your day, but not in the direction you want. You end up somewhere. It's good that you say people can't relate. But a lot of people use Codex and Claude to build dashboards for their e-commerce shops and things…

…and then I just want to offer a hint: make sure what you want done gets done. If you're aware that AI will most likely try to lead you down side tracks, because perhaps everything matters equally to it and it doesn't understand the priority…

…then at least you can reduce the likelihood. I buy that. Okay, enough about AI. What are you most looking forward to this autumn? One thing happening now is a lot of events.

You've probably got some new gigs coming up. What are you most looking forward to? I've been so focused on the next few weeks that I actually don't know. How about you?

I've spent the summer reflecting on what I want to spend my time on. For one thing, I'm looking forward to all the e-commerce events. I'm going to Pindex, or sorry, I've just been there.

Now, on 2 September, I have Resumé's tech event, Breakit's tech event and ehandel.se's tech or e-commerce event. There's Björn Påhlman's e-commerce event, whose name I've forgotten…

XP Awards, which I hope those of you listening will sign up for. Then we have the equivalent of D-Congress in Stockholm, which is usually much smaller and held in late October or November. There are quite a few events, and it's always fun to meet colleagues and industry colleagues and have a chat. Probably a lot about Meta and AI, because that's where things are happening.

Then we have lots of great new hires at Såpa. I'm looking forward to building the company there. We've just landed our Head of Retail, who started last week. We have a CFO role out that we've been headhunting for.

It's company-building like nothing else, to stabilise, build and then be able to put our foot on the accelerator again. I'm really looking forward to that. A lot is happening at Porobic Group too. We have our kickoff next week and are going to Crete together to brainstorm new ideas ahead of Black Week and do a lot of work on that for our brands. We're going to Euphoria Resort, a bloody lovely place, but the focus is, of course, on workshops, work, team-building and getting the whole gang together. Some of us are in Sweden, and some in Europe, spread across different countries. We only meet in person once a year.

A fun thing; I'm looking forward to it. There's a lot happening soon. We'll soon be talking more about Black Week, Black Month or whatever it's called.

Black Weeks. Black Week Month. Once I named a campaign Black Month for a Swedish brand selling in the US. We got so much shit in the comments: ‘But it's not called Black Month. Do absolutely not. This is racist,’ people called us. I was like: we have to take that ad down.

It's controversial to call it Black Month. Actually—we won't get into it now—Black Week itself is quite a dark name when you read up on it. They've tried to change it several times.

Oh, I didn't know that. In any case, it started with Black Friday. Then it became Black Week, now Black Weeks in the plural and then Black Friday Month. A few different names. Either way, we'll be talking quite a lot about it because we're getting started. We begin lead generation fairly soon and have some activations ahead of Black Week and Black Friday. It's the highlight of the autumn for a lot of e-commerce businesses. Fun on the consumer side too: you find things to buy and start writing shopping lists for the home to buy during Black Week.

That will be an exciting episode, of course. Something completely different, since we've talked a lot about AI: I invested in another company, which we teased before the summer, that works with posters.

Do you generate the prints with AI? Yes. We've built an entire workflow. We've looked at the fact that most poster companies generate prints or posters with AI. The vast majority do, even the ones saying it's limited edition and so on. We have a very small team of inexperienced e-commerce operators, but very driven people. So I thought…

When they pitched the idea of a poster company, I thought: not another one. How do you do it better than anything else on the market? But they're passionate people at heart and have their own print shop outside Gothenburg…

…where they practically live and work a great deal. They were open to the AI wave and could adapt the business. Posters have good margins, but all the players in the market offer very steep discounts today. Over the summer, we tried to see how we could build a completely automated workflow.

We have an AI that scrapes the internet for trending posters, recreates them and creates similar posters. Then it uploads them to Meta, to the Meta account, as video ads and still images. It writes copy for both the product pages and the ads, including the ad text above them.

It puts everything in draft. When I wake up, I can see that ten posters were made overnight and the drafts are in the Ad Library. I just go in and approve them: tick, tick, tick. Then I have new items on the site that I'm also testing in ads, to see quickly whether the posters are a hit or a miss.

Then it's also supposed to analyse. We've built our own dashboard that goes into Meta and looks at metrics such as hook rate, hold rate, CPM, CTR, CPC, ROAS and a few others. It weights them differently depending on whether it's a video, still image, carousel or whatever type of ad and creative it is.

It can flag things: if it's green, it's a good ad. If it's red, you should pause it. Eventually, you want AI to pause the ad too, based on very specific criteria.

You can also say that ads should spend a certain number of kronor before they're paused. We're working on that. After 48 hours, I'm not joking, the guys come back and say: ‘Now we've finished most of what we've talked about.’ I'd set out bullet points that I thought would take two months. Forty-eight hours later, they've built the whole dashboard and the workflow that does everything I mentioned.

I was like: fucking hell, that was fast. Initially, the company was turning over a couple of hundred thousand, 200,000–300,000 in May. In June, we were able to increase quite significantly, particularly towards the end, before really taking off in July and reaching around 1.5–2 million. Now August is behind us, and we've already beaten July's revenue and continued trending upwards. We're almost at 2.5–3 million. Really, the season is ahead of us as we move into darker times, with Black Week and Christmas shopping. It's all ahead of us. It's the company's first year of operations, so it'll be fun to see…

…what we close the financial year at. Given how driven the guys are and all the AI automation, I don't think we'll need to hire that many people, but we could still turn over 30, 40 or 50 million on a rolling twelve-month basis. That's pretty cool.

It's fucking amazing. I also think we really need to show how to do all these things.

Absolutely. We fucking need an episode with Oliver, who's on the founding team and has been the key person behind it. I come up with good ideas for how we should build the process and the strategy behind it. He's sat with the team and built the entire workflow…

…set up and reviewed the Meta processes. It's extremely interesting. It takes some time, but it's almost an AI sandbox for me. I pass all my ideas and creative outlets on to the team. They build so quickly, and it's fun to see it become reality. Of course, there are bugs and things that don't work. The system gets so much bloody data coming in that it crashes and you have to reload. So far, there's a lot of that, but…

…we're well on our way towards it being almost automated. We've made quite a few changes to the site and are moving in the right direction. Such as? Among other things, there were perhaps 5,000–10,000 posters when I came on board. I thought it was difficult and disjointed, with no common thread. We're trying to create one. It's still disjointed, with lots of categories, but I'm trying to cut things.

We can see that the system can automatically flag posters that aren't selling. If a particular poster hasn't sold anything in five days, it goes into a segment we call ‘Final call’, something like that.

Is that connected to Shopify's MCP or Facebook's MCP when you decide that? Shopify's. So you check Shopify data for how many orders have actually been sold…

…because with Facebook's MCP, you're checking what we've shown and what Facebook says we've sold. Sometimes they can be completely disconnected. Facebook might think we've sold 50 of one we've sold three of.

If we've sold three, it gets to stay. But if we've sold zero over the past few days or weeks, perhaps it should be phased out. Because so many new posters come in every day, we also need to phase out at least as many. How do you quality-check the material…

…so you don't have a poster where the eyes are a bit way off? That's still human. We have a creative director behind the screens who looks at it. Even if ten posters are ready for publication, scraped from different things trending online. That doesn't have to mean it has scraped other poster companies. We do that a little, but also not really.

We look at what trends are out there right now. On Google Trends, for example, we see that the Swedish election next week is trending. It's huge. We know there are parliamentary and municipal elections. The whole TikTok feed, social feeds and media landscape are talking about it. Should we develop posters about the election?

Do you make parodies of politicians then? Maybe. Why not? We can Americanise it a little. We can do other funny memes that turn up, or if someone makes a blunder, we make a poster out of it. It becomes a fun thing that's contemporary and current.

How do rights work when it's another person? It doesn't have to be another person. Maybe we just have a saying from someone, or a caricature…

Exactly, a caricature of that person. Then I don't know the legal side. Maybe you can have another person too. I don't know. Perhaps the copyrighted material is your own. I don't know exactly how it works…

…because they are public figures, after all. It's hard to say where the legal boundary is. That's why we have Daniel Dernhagen on the founding team, the lawyer I've used…

…over the past few years. Daniel is involved in all the companies I invest in. It's fun, and he can handle the legal side and everything you mentioned. The most important thing is that I don't have to use my mental space on: what about the law? Is it okay or not? We have a group with Daniel where he makes the call.

Then you can throw out the ideas. Exactly, he becomes a filter. But it's fun to see how far we've come in a short time and how revenue is surging. The goal has been… The founding team see it as the first billion-krona company. They sent me a poster saying ‘Zero to a billion’ before we'd even signed. I thought those were very tough ambitions, because I know how small the market is in Sweden. Then we have to go beyond the borders. Right now, we have a local print shop in Gothenburg. Have we thought about that? Most poster companies, if we take Andy Okay or something similar that we've talked about before on the podcast, sell…

…a great deal outside the country, specifically in the US. How is a local print shop in Gothenburg going to ship to the US? It doesn't work. Then perhaps we have to start doing print-on-demand, but that goes a bit against the company's vision of high quality and quality-assured material that we produce locally. It's difficult. We'll see.

Then it's more of a logistics and supplier issue. A hundred per cent. In all likelihood, we'll solve that when we take that step. Right now, the goal is within Sweden's borders. It's fun to see that it works…

…and that the market definitely isn't saturated if you do the right things. An ongoing story on the podcast. It'll be fun to follow, and to have Oliver on to show in more detail how it's done. We'll have to invite him up. Oliver, I know you listen to the podcast, because you always tend to comment as soon as it comes out. Now you can hear that Fredrik wants you on. We'll have to invite you next time you're in Stockholm.

Then we'll share the screen and show it. A hundred per cent. But let's not drag this episode out. Is there any final thing you've been thinking about that you'd like to add?

If you have questions or thoughts, write to both of us on LinkedIn or comment on any platform. We try to answer as best we can. It's fun to be back. I was going to say the morning sofa with you, Fredrik, but now it's the afternoon sofa in the podcast studio, with you and everyone listening who's stayed with us throughout the transition period for Noll till 100. The podcast is evolving and getting stronger. If you have suggestions for guests or topics, write to us on the various platforms, as Fredrik said.

We'll get better, I promise. We have a crazy number of plans for the podcast. It'll be fun to tell you as we get closer. The final sign-off is, as usual: if you like the podcast, recommend it and share it with colleagues, friends and acquaintances. Write a review or a comment on the platform you listen on. Especially Spotify, where we now have the most listeners, but also YouTube, where we publish full-length episodes so you can also see us talking on the sofa.

Thank you so much for listening. We'll be back next week, Thursday as usual. Bye! Yeah.