Skip to content

AI is perfect for service businesses

· Martin Thorborg

AI in service businesses can make traditional companies more efficient. I talk about Shup, Amalo, RenBåd and Dinero’s support – and the opportunities for new entrepreneurs.

Episode 127 · 23. April 2026

Listen to the episode

Watch the video episode

Work comes along to France

Hi, and welcome to Thorborgs dagbog. It’s 23 April. My episode schedule is a bit up and down. Sometimes I’m inspired; other times I simply forget. I’ve also been travelling quite a bit.

Instead of an ordinary day working from home, I sometimes spend a couple of days at our flat in the south of France and add a weekend. That gives me a few days for long walks, thinking, planning and clearing my head. A lot of my work is strategic right now, especially around artificial intelligence.

Tonight, I’m going on Debatten to discuss how AI might affect our society in the near future. Apart from that, I have meetings, journalists and board work. I’ve joined the board of Fiscozen in Milan, which is also part of Visma. It turns out to need a little more hands-on work than I’d originally expected. That’s how it goes sometimes.

But four days in France still do something for me. I think it’s a fantastic way to clear my head.

A proud father at Høvelte Barracks

I’ve also sent my son Mads off to the Royal Life Guards. It’s a big moment as a father. I don’t cry very often, but I certainly came close.

Seeing your grown-up son walk through the gates of Høvelte Barracks, where I started around 35 years earlier, is a special experience. The same place and in some ways the same task: first training as a soldier and later guard duty.

Many people see the guards at Amalienborg as mainly decorative. But it is also a real responsibility and training that you take seriously. Now Mads sends pictures of his locker, his camouflaged face and life in there. The locker is certainly tidier than the one at home usually is. It is bloody lovely following along. I talk about my own time in the Royal Life Guards in If You’re Gonna Be Dumb, You Gotta Be Tough.

I’m inspired by the people around me

I spend many hours reading about AI and following what’s happening. But one of the greatest privileges is working with people who are much sharper on the technical side than I am.

We’ve just spent a few days in Sweden with the whole of Dinero, around a hundred people. Some of our colleagues work directly with AI. I talk a lot with David Andersen and Rasmus Madsen about strategy, and with our CTO and others about the technical possibilities. Where is the world heading, and what does it mean for us?

It also becomes concrete when I look at the smaller businesses I’ve invested in. There are enormous opportunities to make traditional businesses more efficient while giving customers a better experience.

Shup makes starting an online shop easier

Take Shup, which makes an online shop system for small businesses. My sister Katrine runs Katrines Have, and selling tomato seeds is one of the things she makes a living from. She sells a lot in spring and continually works on new products and other types of seeds.

It’s a typical small business that needs to get started quickly without a big technical project. With Shup, she can create her online shop and build on it at her own pace.

AI can help with small things too. Upload a product picture and you can get a suggested description. If it’s a red teddy bear, the system can give you a first draft to adjust. It doesn’t have to be perfect first time to be useful. You have something to work from instead of an empty field.

That’s the sort of solution I like: make it easy for the customer to get started and spend their time on the actual business.

Amalo has automated its administration

At Amalo we’ve changed managing director. Rasmus needed a change and is still a co-owner. Christian, who was previously CTO, wanted to take over.

I told him honestly that I was a little sceptical at first. Was he too much of a geek for that role? But it’s turned out to be an enormous gift. He looks after the business really well alongside the employees, and his technical background creates some very special opportunities.

At the time of recording, he’s built several AI agents that handle different parts of the office work. From the customer ordering a heat pump to delivery and installation, much of the process is automated.

Quoting is an obvious place to start. When a particular heat pump is ordered, you need to know which parts are required, what the installer needs to bring and how to plan the job. Pipes, fittings, cables and other things. I’m no heat pump expert, but I can understand how much time it saves when that doesn’t need handling manually every time.

There’s automation around marketing and monitoring ads too. That means a few people can handle a large amount of administration. Those savings can help make us both cheaper for the customer and more profitable.

AI gives us a first look at the boat

At RenBåd Rasmus Madsen has built a solution I find really exciting. We look after a large number of boats, and when a boat comes in for winter storage, we take pictures all around it.

The pictures can be used to alert the customer to visible damage and give an initial indication of what should be inspected or repaired. That might be damage to the gelcoat, for example. If nothing is done about it, in some cases it can lead to bigger, more expensive damage.

That doesn’t mean a picture can reveal every fault or that AI replaces a proper inspection by a mechanic. But it can help find things that are easily overlooked and point to what our professionals should examine more closely.

We’re also working on letting customers send in their own pictures and get a preliminary quote based on what’s visible. At the time of recording, we’re almost finished with that solution and have had good results in tests.

When we draw people’s attention to something, many choose to have it fixed. In the tests I talk about in the episode, around half the relevant customers said yes. Of course, it’s an opportunity for additional sales, but it’s also a service. The customer is told there may be something that could develop into an expensive problem.

You can say no, get someone else to fix it or do it yourself. It’s entirely up to the customer. But many would like us to take care of it while the boat is with us anyway.

Convenience is a big opportunity

I think convenience is one of the big trends. People want to spend their free time doing what they love and pay someone to take care of some of the things they find troublesome.

Sandblasting and antifouling your boat might not be your idea of the world’s best weekend activity. Perhaps you’d rather go sailing. If we can handle storage, transport, servicing and preparation properly, owning a boat becomes a different experience.

There’s an old saying that the best days in a boat owner’s life are the day they buy the boat and the day they sell it. That’s part of the experience we’re trying to change.

Our idea with RenBåd was to take a traditional industry and digitise more of the work around it. The skilled tradesperson still needs to be there. But if he spends most of his day on the phone, quoting and planning, he doesn’t spend enough time on what he’s actually good at.

We’ve also had lots of strategic discussions: what kind of business are we? Should we handle major insurance claims, sell boats or focus on particular jobs? Using technology hasn’t made it simple. But now we’ve got much of the foundation in place and can keep working on scaling.

An agent works while you do something else

There’s a difference between going into a chatbot to ask a question and having an agent set up to follow a process. For example, an agent can monitor incoming enquiries, gather information and draft a quote.

At first, you can have it prepare the reply for you to approve yourself. Then you see whether the language is right and whether the quote is correct. As the setup improves, you can automate more of the steps where it’s sensible to do so.

I imagine that, in time, administration in businesses such as Amalo and RenBåd could be handled by very few people. Perhaps one person keeping an overview, plus salespeople or other roles where they add value. That’s my ambition and vision for the future, not a claim that all businesses can already operate that way.

The interesting thing is that more customers don’t necessarily require proportionally more people in the office. If the processes work, you can hire more tradespeople and do more work while the systems handle much of the additional administration.

Dinero support is a concrete example

At Dinero, we froze hiring in support around two years ago. At the time of this episode, we have a support team in the low forties. Without AI and automation, my estimate is we’d have had significantly more.

AI handles a large proportion of enquiries, and customers’ ratings of the answers are close to their ratings of human replies. Customers can also get help at night and receive an immediate answer, rather than waiting until the next morning and then waiting for a support person to become available.

For some customers, it’s also nice not to have to write to a person. Perhaps they type slowly or worry about spelling mistakes. People have different preferences for getting help.

My estimate in the recording is that support could have around 35–38 people by the turn of the year, whereas without the changes we might have had around 60–65. That’s an expectation, not a finished result.

Across Dinero, we reached around 115 employees and are now around 105. My guess is that we could be around 95 by the end of the year, whereas without AI we might have needed around 140. Again, those figures are my estimates at the time of recording.

AI isn’t free either. Licences, development and operations cost money. So you can’t just take the difference in staff numbers and say the whole amount goes straight into your pocket. Right now, it’s primarily about building the business we believe the future requires.

If we don’t develop, our competitors will

I also get criticised for cutting jobs. I understand that these developments can be unsettling, but I’m responsible for keeping the business competitive.

If I can’t figure it out, Visma can find another managing director. And if the business doesn’t keep up, customers can go to someone who’s done it better. Then we don’t just lose some support tasks. We risk losing the whole business.

My job isn’t to keep as many people as possible busy with manual work if it can be done better another way. My job is to build a good, lasting business that delivers value to customers.

If you think a business should be organised differently, you’re very welcome to build it and show that the model works. I get sceptical when a particular ownership structure or organisational model is presented as a guarantee of success. You still need customers, good processes and sound finances.

Henrik built two solutions himself

A fun example is my friend Henrik Stender. I’ve never seen him as a big IT guy, even though he’s very smart. He hasn’t spent years programming.

Yet in a short time, he’s built two solutions with AI. One was a booking system with SMS reminders for the hairdresser who comes to us. The other was a solution for sharing pictures from the trip to Sweden.

I went in, signed up and uploaded my pictures. It worked, and I thought it was a great service I hadn’t heard of. Then he told me he’d made it himself.

A few years ago, I think it would have been a completely different project, requiring a capable programmer and much more time. Now he’d built something useful in a few working days, including the interface, database and integrations. That doesn’t mean every project is finished and secure in a couple of days, but it shows how much easier trying out an idea has become.

The window cleaner can save the trip to give a quote

Imagine what that could mean for a window-cleaning business. The customer enters their address, and a system uses available information, pictures and the customer’s own answers to give an initial estimate.

How many windows are there? How high are they? Is special equipment needed? The system can’t know everything for certain, but perhaps you can quote for the first visit and then adjust the regular price once you’ve been to the property.

Compare that with the usual process: the customer calls, you agree a time, drive there, walk around measuring up, perhaps have a coffee, drive home and write the quote. It can easily take a couple of hours when you include everything.

As a hypothetical example, if we value the time at 400 kroner an hour, that’s 800 kroner per quote. If only one in four customers says yes, you’ve spent time worth 3,200 kroner to acquire one customer. That’s before the actual work and all the other costs.

It’s a rough calculation, but it shows how much making quoting more efficient matters. At Amalo, we’re working with an estimated saving of around 10,000 kroner per customer by automating that process. Those savings can be shared between the customer and the business or used to become more competitive. If it holds up, for example, we can give the customer a 5,000-kroner discount and keep the other 5,000 kroner in the business. That’s an estimate at the time of recording.

AI in service businesses: An open opportunity in traditional industries

If you want to be an entrepreneur, I think traditional service businesses are an exciting place to look. Not just at the actual trade, but at everything around it that takes time and makes the customer’s experience awkward.

It reminds me a little of having a shop on Falkoner Allé in the early nineties. I had to be in the shop to sell. Today, a small business can set up an online shop and sell while the owner does something else. That changes the opportunities.

The same can happen with administration in a service business. You still need to clean the windows or install the heat pump. But perhaps you don’t need to take every call, write every email, plan every job and spend a large part of the day giving quotes yourself.

That’s the opportunity I find so fucking exciting. Get technology to handle some of the hassle so people can spend their time on what customers actually pay for. And build it so more customers make the business bigger without making administration proportionally heavier.

I’m nearly there now. I have a job at Dinero to look after, and I feel really lucky to be surrounded by such talented people and get inspired. I also take that knowledge into the businesses I’ve invested in and into what we’ll do in the future. Perhaps I’ll see you on Debatten tonight. Bye!

See more episodes of Thorborgs dagbog in writing →