Field notes from
inside the work.

I didn't want to become a PowerPoint engineer.

The typical leadership path goes one way: more meetings, more dashboards, more decisions made at a distance from the actual work. I went the other way.

So this year I went hands-on deliberately. Visual Studio Code, the command line, GitHub, prompt engineering. Building skills, feeding the system with the right context so it actually gives useful output. Tools I knew but hadn't worked with deeply enough.

I wanted to experience the real friction our software engineers navigate every day. Not hear about it, not read about it. Feel it.

What you discover when you work this way can't come from a status report.

The best panels go off script. Ours did, and we landed on the question underneath all the others: what are we actually racing for?

This week at Almedalen I sat on two panels, both asking whether Europe can close the AI gap.

On the Global AI Innovation Race panel, the surprise was how little we disagreed. I sat with Alexander Kvist, Shreerang Talekar and Erik Slottner, four of us meeting AI from research, enterprise scale, policy and industry, and we kept landing in the same place: Europe doesn't lose this race on talent or values. It loses on how fast we turn pilots into production.

The next day's roundtable, with energy leaders Carin Kuylenstierna, Erik Persson and Jörgen Hasselström, stayed on the harder question: how you get from experiments to projects with a real return. The moment that stuck came when the moderator asked whether an AI-native company could make us obsolete. My answer: it is not that simple. There is a difference between AI-first and AI-native. None of us at that table is AI-native, but all of us are pushing to become AI-first, and the AI-native challengers have the models without the products, the manufacturing, or the decades of learning to build them at scale.

Thank you to Linda Nyberg for steering us somewhere unexpected, to Fredrik Silfver and Jonas Åhlen at Eidra for hosting, and to everyone who shared the stage and the table.

So what are we racing for? My answer is not to avoid losing. It is to protect the life we have built in Europe, and to build a better one, for this generation and the next.

In 2021, we were reviewing our portfolio, competences, capabilities, where we were heading. The question that kept coming up: how does a team like ours stay cutting edge?

That conversation became our AI initiative. What followed wasn't a transformation program. It was structured experimentation. A small core team with a clear mandate: collect ideas from anyone in the organization, evaluate them, build a proof of concept, then an MVP, then scale. Two-week cycles. If it didn't fly, scrap it and start the next one. By 2023 we formalized this as SWACC, a dedicated software accelerator team, to keep the flywheel turning.

Not every idea made it. Some would only ever benefit a single user and couldn't scale. Others integrated directly into CI/CD pipelines and ran for everyone. One of them, KnutGPT, could analyze hundreds of log files and surface root causes automatically. Work that previously meant experts reading through millions of lines of output.

But the technology was never the hardest part.

My belief: 20% is tech, 80% is people. This was change management as much as engineering. People needed to get familiar with something unfamiliar. The more they experimented, the more they accepted it.

With the ChatGPT moment, we were building LLM-based solutions with RAG all the way to agentic systems. Not because of a grand plan, because the team wanted to try, and there was a problem worth solving.

The technology gap closes fast. The culture gap doesn't.

Most people think of cybersecurity as a shield. 🛡️ Something you add around a product to protect it.

We think about it differently.

When the security work started in Lund, we brought something unique with us: deep expertise from the telecom world. Mobile devices, secure production environments, protecting content on hardware that's always connected. That knowledge transferred directly into automotive, and the result was ESCRYPT CycurIDS, an intrusion detection system for CAN and Ethernet networks, and ESCRYPT CycurGATE, an automotive ethernet firewall.
Both are products of ETAS, a Bosch subsidiary, now used by major OEMs building software-defined vehicles (SDV).

The philosophy we carried into that work has stayed constant. Security isn't a layer you add at the end. It's a property that either exists throughout the system or doesn't exist at all. Inside-out, from the start.

That matters more now than it did in 2016. What needs protecting has changed: it's no longer just personal data on a device. It's AI systems, machine learning models, connected infrastructure. The attack surface keeps growing. The inside-out approach isn't optional anymore.

The most expensive security decision we see companies make is the same one, every time: deciding to address it later.

Ten years of this work, built by a team that treats security as a craft.

Most teams build the product first and the infrastructure later.

We did it the other way around.

When the Bosch R&D Center in Lund opened in 2016, one of the first decisions we made was to treat CI/CD not as a tooling choice but as a cultural commitment. Not as an IT standard. As a way of thinking about software development itself.

Setting the bar higher than the norm and refusing to let "it works" be the same thing as "it's good."

What followed didn't always look impressive from the outside: pipelines, test automation, on-prem to cloud migration, centralized tooling. The kind of infrastructure work that earns no applause. But it's also the foundational work that made everything else possible: the Android migration, the ADAS data loops, and eventually the environment that let our team walk into a Google Bootcamp in Sunnyvale and demonstrate working SDV implementations on series-production hardware, not slides.

The platforms Lund built became a reference for other Bosch units. We didn't just adopt DevOps. We trained organizations that were older and larger than us on what it looked like in practice.

Ten years in, AI-first development and software-defined vehicles run on that same foundation.

That continuity is intentional.

What does an engineering culture look like when it treats infrastructure as the product? Curious what you've seen work, or not, in your teams.

Recently I listened to a podcast about joy versus happiness. There's a difference, and it matters.

Happiness comes from the Norse word hap - happenstance. It's the accumulated sense that things are going your way. It requires luck to add up. Most days, it doesn't.

Joy is different. Joy doesn't wait for the conditions to be right. It surprises you in the middle of the hard thing. It's the soul's yes, where, despite everything, it feels good to be alive. It goes deeper and lasts longer.

I've been in leadership long enough to experience both.

Happiness looks like a strong quarter, a product launch that went well, a presentation that landed. Real moments. But they don't sustain anything.

Joy looks different. For me, it shows up when a team does something that surprises even themselves. When a problem that seemed stuck suddenly cracks open. When I'm building something that didn't exist before.

For me, I also know what takes it away. Bureaucracy for its own sake. Long stretches with no room to initiate anything new. And when joy disappears, we fill the gap with manufactured happiness, scrolling, comparing, anything to feel something positive. None of it works.

The podcast ended with a story about a cathedral in Portugal. Centuries of artisans embellishing every wall, adding pineapples, more pineapples, endless ornamentation… and nobody ever stopped to build the ceiling. It stood open to the sky not because it was designed that way, but because everyone was too busy adding detail to finish the building.

It's easy to keep adding pineapples.

Worth asking yourself this week: is what you're doing bringing joy, or just the next layer of ornamentation?

Ten years ago, a small group of ten people got together in a living room near Lund.

No office. No infrastructure. Just a handful of engineers, ready to build something new and exciting, and probably too much coffee. ☕

That was how the Bosch R&D Center in Lund began.

I wasn't there on day one. I arrived later, and in 2020 I took on the responsibility of leading what that living room had become. Honestly, when I did, I wasn't entirely sure I was ready.
Leading an R&D center through a global pandemic, in a country that wasn't my own, with a team I was still getting to know — that's a humbling place to start.
But here's what I've learned about building things: you're never really ready. You start anyway. And with a team like this, I never was alone.

Today, that living room has grown into a center of more than 200 people. We've developed award-winning products, patented new solutions, and become a home base for around a hundred master's theses. We've launched projects and initiatives that continue to push us forward — and we're just getting started.

What I'm most proud of though isn't any single technology. It's the culture.
The fact that people from dozens of nationalities chose to build their careers here, in Lund, together.
The endless push for excellence and the continued hunger for learning and growing.
That we argue about the right technical decisions because we care — and then we move forward together.

I also want to take a moment to recognize where we are. Lund is an incredible place — a unique ecosystem powered by tradition, innovation, and talent. Being located close to Lund University, The Faculty of Engineering at Lund University (LTH), Medicon Village, and Ideon Science Park isn't a coincidence. These partnerships have shaped who we are.

And yet — every week I meet people who are genuinely surprised when I tell them what we actually work on at Bosch Lund.
That's something I want to change.

Ten years is worth pausing for. Here's to the next ten.