Our trip to AI Week in Milan
When you gather thousands of techies, global giants like Google and Microsoft, and some of the world's sharpest minds in artificial intelligence, you expect a completely streamlined, futuristic experience.
And it was. Up until it wasn't anymore.
Our CEO, Jeppe, and CTO, Vadym, have just spent a few whirlwind days at AI Week in Milan. They have returned with their bags full of Italian coffee impressions, lots of pictures, and — most importantly — some healthy, down-to-earth reflections on where the AI world is actually heading right now.
Here is a short account of what they experienced, and what we are bringing home to our everyday work.
Between Google speakers and Italian speeches
The conference was massive, and the energy was absolutely spot on. But amidst all the talk about future algorithms, we quickly realized that technology still has its charming limitations. More than half of the presentations were delivered in Italian. Fully in the spirit of the times, the organizers had relied on an AI-powered live translation within the conference app, but let's just say the artificial intelligence went into overdrive and didn't always hit the mark.
Combine that with food lines that required some very analogue patience, and we were quickly reminded that the robots haven't completely taken over just yet.
But up on the stages, there was gold to be found. Two presentations in particular got the conversation flowing over dinner.
Data beats hype — every time
On day two, Jeppe listened to a presentation by Marinela Profi with the catchy title: "Why leaders who win the Agentic AI race won't be the ones who move fastest."
In the executive boardrooms, it's often about being first. But Marinela's point was the exact opposite: It's not about running fastest after the latest language model, which changes from week to week anyway. It's about strategic patience — and about building a proper, structured data foundation first.
She pointed out three classic AI pitfalls that we can certainly relate to:
-
To blinding oneself to the technology itself instead of focusing on the business.
-
To choosing speed over a well-thought-out strategy.
-
To forgetting to evaluate when AI even makes sense — and when traditional methods are actually far better.
In short: AI is worth nothing without proper data. And you shouldn't use AI just because it's a trend.
We must control AI — not the other way around
As our CTO, Vadym had his antennae out for how all of this affects the actual craft behind software development. He took particular note of the session "Your brain on ChatGPT" — and brought home the following takeaway:
AI is an excellent tool for optimizing processes and automating routine tasks. But there is a downside if we let go of the reins entirely. If we blindly let AI drive our processes, we risk weakening both our own cognition and our developers' deep, structural understanding of the systems we build.
That is why we return home with a clear philosophy: We use AI to optimize — but we retain full control and understanding. On critical core systems, we need predictable automation, not unpredictable AI coding. The human sits in the driver's seat, so we actually know what is happening if — or when — something fails.
Back home in Denmark
After a couple of days with our heads planted firmly in the future, it was time to head back to Denmark. But as if to emphasize that digital infrastructure sometimes has its own sense of timing, the latest Nets outage naturally hit right during our journey home. It caused a few logistical challenges and a bit of a sweat at the checkout counter — putting an almost perfectly ironic bow on the whole trip.
We have landed back in our everyday routine, full of impressions, pizza, and coffee — and with a slightly thicker skin against AI hype. Milan reminded us of something important: the most interesting conversations didn't happen from the stages, but over dinner afterwards. And that is probably where the best decisions have always been made.
Thank you for the trip, Milan.