Post by Michael Nørklit

The Compass Problem: What AI Really Demands from Leaders In 2011, Marc Andreessen wrote that software was eating the world. He was right. Now AI is eating the software and the boardroom hasn't caught up. I've spent 13 years building a technology business in the Nordics. I watched perpetual licences give way to SaaS, then cloud displace on-premise. Now I'm watching AI agents begin to automate the tasks that justified the software in the first place. Each wave faster than the last. But the conversation I keep having with customers, with investors, with other leaders isn't really about the technology. It's about what leaders are supposed to do when the technology is moving faster than the governance around it. Here's the problem: as AI gets deployed team by team, local optimisation accelerates. Individual units get smarter, faster, leaner. But no one is asking whether all those local wins are pulling in the same direction. In a decentralised AI environment, the compass problem gets harder, not easier. Leadership used to be about setting direction and allocating resources. AI changes the resource equation but not the direction problem. If anything, it makes it more urgent. When decisions can be made in milliseconds by an algorithm, which direction becomes the only question that matters. Let me make this concrete. For the past several months I've been working directly with Claude AI not as an experiment, but as part of how I draft, pressure-test models, and structure my thinking before a meeting. What surprises me isn't the speed. It's how mercilessly the tool exposes the quality of the question you bring to it. Vague brief, vague output. Sharp brief, sharp output. For senior leaders, that's the uncomfortable insight: AI doesn't replace judgement, it amplifies whatever judgement you already had gaps included. And the trajectory is steepening. We're moving from AI as a tool you query to AI as agents that act on your behalf across systems, vendors, and timeframes. It is no longer enough that your people know what to do; you now have to articulate intent clearly enough that an agent can execute it without you in the room. Most organisations cannot do this yet. Most leaders have never been trained to. There's a second pressure that rarely makes the leadership agenda: the energy cost of intelligence. We are building data centres at a pace that quietly undermines the sustainability commitments in the same annual reports celebrating AI transformation. The clean, frictionless image of AI doesn't survive contact with the electricity bills. The leaders who navigate this well won't be the fastest. They'll be the ones who hold two things at once: the speed and ambition AI enables, and the clarity of purpose that stops an organisation from optimising itself in ten directions at once. That's the paradigm shift. Not AI as a tool most leaders have accepted that already. But AI as a stress test of whether your leadership actually knows what it's for.