Edition #5: The Cost of a Frictionless Ecosystem

There's a moment in every maturing ecosystem when things start to feel effortless. Everyone speaks the same language, meetings run smoothly, nobody pushes back.

That's not health. That's the beginning of decline.

It is called harmony bias: the tendency, as an ecosystem matures, to keep curating for agreement: the same market view, the same technical assumptions, the same instincts. It feels efficient. It's actually the moment an ecosystem stops learning.

Real progress doesn't come from consensus. It comes from friction. After years of building this Campus, I've come to think there are two kinds you have to design for on purpose.

The Researcher and the Founder Don't Move at the Same Speed

Let’s talk about vertical friction. This is the tension between people at different points in the AI-building process. The researcher is playing a long game measured in years, not weeks: robustness, peer-reviewed truth. The founder operates on a two-week sprint: ship-speed, market signal. Put them in the same room and the founder gets pulled back toward reality, and the researcher gets pushed to explain why their work matters now, not eventually.

The same tension shows up between policy and engineering. The policymaker builds guardrails against systemic risk. The engineer builds to push boundaries. On their own, each is incomplete. Policy without engineering input is theoretical, engineering without policy input is reckless. In our Town Hall sessions, roundtables and events, that tension between them is the filter that separates novelty from impact.

Your Competitor Is Not Your Problem

This second kind of friction is harder to sit with, because it involves people who are, on paper, your rivals.

I still meet founders who assume the safest move is to keep their distance from anyone building in their vertical, lock the doors, minimise contact, protect the edge. In AI, I've watched that instinct cost people more than it protects them. The stack moves too fast for isolation to work.

What I notice, again and again in our community, is that two founders in the same vertical rarely lose to each other. They lose to the same bottleneck: data-access rights, compute cost, a slow enterprise procurement cycle, unclear regulation. Those are shared problems, not competitive ones. On neutral ground, founders trade notes on exactly those problems without ever touching what actually makes them competitors.

Who's Holding the Room

Both kinds of friction only stay productive under one condition: the platform hosting them has to be genuinely neutral. Not answering to one agenda or optimising for its own extraction.

That's the part of this job I think about the most. A neutral platform can put a researcher and a founder at the same table without either one performing for an investor in the room. It can put two competitors under the same roof because its incentive isn't to referee the competition, it's to keep the whole ecosystem healthy. That's what turns friction into signal instead of conflict.

The takeaway

Stop looking for a ecosystem that validates your assumptions or shields you from your competitors. Look for one that does neither, on purpose.

If nobody from a different discipline is pushing back on you, and you've never had to share a room with someone building the exact thing you're building, you're not moving fast. You're just moving comfortably. Those aren't the same thing.

‍

---

‍

The "AI Ecosystem Playbook" series goes behind the scenes of how Europe's densest AI ecosystem actually gets built and run. Written by Sif Björnsdottir, Head of Community & Growth at the Merantix AI Campus in Berlin, the series shows the systems and decisions behind a functioning, multistakeholder AI hub — and what it means for the founders, engineers, investors, and corporates deciding where to plug in next.