From Uncertainty to Decision

Over two decades working with boards of directors, I learned something that changed the way I think about innovation — and that I now apply directly to AI:

The problem is never a lack of data. The problem is a lack of causality.

Most organisations accumulate massive volumes of data and organise them into categories of correlation. They know what happened, but they don't know why it happened. And without understanding causality, every decision is a guess disguised as analysis.

The concept of Jobs to Be Done is perhaps the most powerful tool I know for finding causality where others see only correlation: don't ask what the customer buys — ask what progress they are trying to make, in what circumstance, and why current alternatives fail.

Apply this logic to AI. When an organisation says 'we need a chatbot' or 'let's implement generative AI', it is starting with the solution. The right question is not 'What AI technology should we buy?' The right question is: 'What specific operational decision do we want to make better — and why aren't we making it well today?'

This reframing changes everything. We stop evaluating AI by its technical sophistication and start evaluating it by the decision it informs. That distance between a model that works in the lab and a model that generates value in production is not technical. It is strategic.

But identifying the right decision is only the first step. The second is managing uncertainty: instead of starting from optimistic projections and investing heavily in the hope that they are correct, we should start from the financial outcome we need to achieve and work backwards, identifying all the assumptions that need to hold true. Then, test the most critical assumptions first.

Applied to AI, this is transformative. Instead of launching a €2 million project and praying, you define: 'for this project to generate return, we need A, B and C to be true. Let's test A in the first week, B in the first month, and only if both are confirmed do we invest to test C.' It is the opposite of AI tourism. It is strategic exploration with decision checkpoints.

AI does not automate tasks — it reconfigures entire competitive systems. The question is not 'how do I automate this task?' but 'how does AI change the way work is coordinated in my sector?'

There is a consequence of this reconfiguration that has become central to my thesis: when AI dramatically increases the supply of answers and insights, economic value migrates to two human capabilities — the ability to formulate the right questions before seeking answers, and the ability to select which answers deserve action. Knowing what to ask and knowing what to do with the answer becomes the most undervalued competitive advantage of our time.

My central thesis is this: the difference between organisations that extract real value from AI and those stuck in tourism is not technological. It is the ability to create a strategic system that transforms uncertainty into decision.

This system is what I teach. It is what I implement. And it is what separates organisations that do AI tourism from those that generate return.