You have been building companies inside DevCode since 2007. When did the model click into being a venture studio?

Honestly, the label came later than the model. We had been operating like a studio for a long time before we used the word. The first three companies, DevCode Payment, Nordkap, and the original consulting business, were built with the same instinct. Share the engineering bench. Share the customer relationships. Share the operating playbooks. The label venture studio entered our vocabulary around 2015. The practice was already a decade old by then.

What is the practical difference between a studio and a fund?

A fund makes a bet and gets out of the way. A studio makes a bet and stays in the room. We co-found, we put group capital and group engineering behind a build, and we operate alongside the founders until the company can stand on its own commercial legs. That has a cost. It also has a return profile that a fund cannot match on the early bets.

Two thousand and seventeen was a big year. DevCode Payment was sold to Bambora, Nordkap to Collector Bank.

It was. And it taught us something we did not fully appreciate going in. The exits were good outcomes individually. But the studio compounding effect came from what we did with the proceeds and, more importantly, what we did with the team and the customer relationships that the exited companies had built.

Meaning?

Meaning that an exit is not the end of the studio value capture on that bet. The engineers go on to build the next company. The customers from the exited company often become the customers of the next one. The lessons compound. That is the studio model working over time.

An exit is not the end of the studio value capture on that bet.

Luminary was sold to White Pearl Technologies last year. Different from the 2017 exits in any way?

The exit logic was the same. But the scale was different. Luminary had grown into an umbrella for over eight thousand specialist consultants across more than eighty firms. That kind of scale changes the studio center of gravity. We are not the same shop we were when we exited PaymentIQ.

You co-founded PayControl this year with Nathan Salisbury. Why now?

Because the autonomous-payments thesis lands now in a way it would not have landed three years ago. The merchants we work with have engineering teams who can run orchestration code. The economics of per-transaction pricing have started to look misaligned. And AI is cheap enough that an in-house orchestration layer beats a vendor pre-trained model on the merchant own data. The thesis works. So DevCode Group and Nathan built it together.

You are also building two more AI-native ventures.

We are. I cannot say much yet. The shape is consistent with how we have always built. Clear thesis, sharp founder, unit economics that do not require a miracle. The difference is that the operating layer is AI-native from day one. That changes what we can do.

What does the studio look like in five years?

I have stopped giving precise five-year answers because the rate at which AI is changing operating layers makes them wrong. What I will say is this. A venture studio that does not internalize AI in its own operating layer, not just in the products it ships, is going to be at a structural disadvantage. We are working on getting that right.

Final question. What do you most regret not building?

Probably the consumer-side identity wallet we talked about in 2018. The thesis was right. The timing was wrong. Someone else built it, and built it well. The discipline of saying no is a real cost. Sometimes the right idea sits in the discipline.

— In conversation with Pedro Hansson, Group CEO