In this edition of Luminar Ventures’ Founder Series, we meet Andreas Lundmark, co‑founder and CEO of Berget AI. After more than a decade building AI capabilities at Boston Consulting Group, he left the partner track to help create a sovereign AI platform in Sweden. In this interview, Andreas talks about why regulated organisations have struggled to adopt generative AI, how Berget found its footing with early customers like the Riksbank, and what it takes to build a resilient company from day one.

You spent many years in strategy consulting and AI at BCG. What made you step off that track and start Berget?

My background is quite classical on paper. I’m an engineer by training, I studied in Linköping and more or less grew up behind a computer. But I left the pure coding path fairly early, moved into telecom and IT, and then ended up at BCG in 2005 where I stayed for nearly two decades.

Over time I gravitated towards the intersection of strategy and technology. We started doing more serious AI work from Paris, sold a very large AI project in the Nordics and I took over that delivery from the tech side. That grew into the Nordic piece of BCG’s AI practice: instead of handing over Excel models, we were delivering real code, products and solutions. It was like building a firm inside the firm - new roles, new delivery models, a global team.

Then life accelerated. I had three children in four years and took parental leave three times, which raised a few eyebrows among international colleagues. 

On top of that I had started to feel the limitations of building products inside a partnership structure. Every krona you put into R&D is, in practice, someone’s bonus you’re reallocating. I enjoyed building really advanced ML based products with real product teams and FDEs, but those products inevitably gravitated towards the biggest markets, so travel would have been a prerequisite but that was not the right time in my life for that.

In parallel I got involved with startups - board work with companies like Validio, some advisory roles - and could see how much faster things moved there. made it increasingly clear: I wanted to build a product and I wanted to build a company. 

When did the specific idea for Berget AI click for you?

Like everyone else, I saw generative AI coming and thought: “This is going to be big.” But seeing the  inside of a large bank during that wave gave a very different perspective.

The bank understood the potential, but within their own rulebook they basically couldn’t do anything meaningful. They were asking: how many years will it take before we’re even allowed to start? Do we have to cheat our own processes to experiment? They could play with non‑sensitive data, but the real value sits in data you absolutely cannot move freely.

At the same time Christian, who would become my co‑founder, was seeing the same thing from his consultancy, especially with public sector clients. They wanted to use AI properly but were blocked by where data would have to go and under which laws it would sit.

It was obvious there was a structural gap. Organisations were being told that AI was the future, while in practice many of them couldn’t safely or legally use the most powerful tools on their most important data. The idea that became Berget was essentially: can we create a platform where they actually can?

For me personally there was also a pull towards building something of my own. As a consultant you more or less start from zero on every new assignment. Here there was a chance to build something that accumulates over time instead of resetting with each project.

How did you and Christian end up deciding to build this together?

We were introduced by a mututal partner. We met, compared notes on what we were seeing in banks, authorities and other regulated environments, and realised pretty quickly that our perspectives lined up.

My initial thought was actually that I might invest a bit, sit on the board, do some advisory work. But the more we dug into the problem, the more operational it became. It felt like something that really needed doing - and that we were quite well positioned to do.

Christian and I spend a lot of time talking about values, integrity and the overall mission, and we consistently ended up on the same side of those issues, both in terms of values and working style.

By that point we had put some of our own money into the company and had already taken a few tough decisions together, which built a lot of trust. From there, we just got going.

You launched quickly and landed demanding early customers. What did those first months teach you?

We launched the version of the service we have out now in late April. Very early on we had the Riksbank as our first customer. That, in itself, was a strong signal that we were solving something real.

One learning was how quickly “pilot” turns into “production” if you solve a real pain. We’ve had partners say they’re just testing a bit, and then a few months later we find out they’re in full production with dozens of organisations on top. One example is an agent platform for municipalities that now serves around sixty organisations on Berget. At that point, you can’t hide behind “we’re just two people in beta mode”. They see you as a production partner, and you have to live up to that.

All of this happened with very little sales from our side. The customers who have found us so far fit very clearly into the buckets we had in mind from the start: public sector, healthcare, education, banks and finance, GDPR‑heavy environments. And their reaction has often been some variation of “thanks for existing - now we can finally get started”. That tells us the need isn’t theoretical.

You’ve talked about not wanting a company that depends on a single “hero” founder. How are you thinking about building resilience from day one?

A pattern you often see in young startups is that one founder becomes very dominant. As long as that person is in every meeting and every detail, things move. The day they’re not around, nothing happens.

From the beginning we’ve tried to design Berget in the opposite direction. It helps that we’re two founders with quite different roles: Christian leads product and tech, and I handle everything else. That split makes it natural for decisions to be distributed. The goal is that people know what they’re supposed to do without a founder having to sit on their shoulder.

We’re also conscious about building structure early. It doesn’t have to be heavy processes, but there needs to be clarity: how we work, what customers can expect from us, what we commit to and what we don’t commit to. Some of that has been “forced” on us by customers moving faster than we expected, which in practice is a good thing. External expectations are a great forcing function.

In the end, resilience is about the company working because the company works - not because two people are running around plugging every hole. That mindset influences how we prioritise hires, how we communicate with customers and even how we think about things like press around a funding round. For us, a round is not a finish line; it’s a starting point that hopefully lets us build more of that long‑term robustness.

What still motivates you most when you walk into the office in the morning?

Right now there are two big sources of energy. One is very concrete: executing on the funding round. That unlocks the ability to build out the team, move faster on the roadmap and be more proactive commercially, which we’re impatient to do.

The other is more everyday and maybe a bit mundane, but it’s what makes this fun. We’re still a very small team. You come in in the morning, you face a set of very real problems from customers in banks, municipalities or other organisations, and by the time you go home you’ve actually built something that solves part of that. It’s a bit like being a carpenter: you leave tired, but you’ve made something tangible.

Customers telling us “now we can finally do this within our own rules” is incredibly motivating. It’s a reminder that behind all the talk about models and infrastructure, this is ultimately about enabling people to do their jobs better in places where it really matters.

With three kids and a fast‑moving startup, do you ever manage to unplug?

At the moment it’s mostly kids and work. The company is very much a passion project, so it doesn’t feel like we’re suffering through it, but there isn’t much space for elaborate hobbies.

That said, I do enjoy playing golf when life allows. It’s pretty much the opposite of the startup rhythm - slow, analogue, repetitive - which is probably why it’s such a good way to reset. For now, though, I’m okay with the handicap going in the wrong direction if it means we get to build something lasting here.

‍

Thank you to Andreas Lundmark for sharing his journey and bold vision with us. For more Founder Series interviews and insights from the Nordic startup ecosystem, visit Luminar Ventures’ News & Insights.