AI STRATEGY
“ And then you need the data, context and monitoring to support it,” he says.“ Historical data can be very useful, but a beautiful model trained on patterns that no longer exist is just a very sophisticated way of predicting the past.”
Learn to love turbulence As a recent report from the consulting firm McKinsey notes, as in previous waves of tech-driven innovation, companies are still falling into the‘ pilot purgatory’ trap, unable to scale their digital or AI strategies beyond initial experiments. Juan José posits that the C-suite is imperative to getting these programmes out of the trap.
Predictive AI inevitably deals with uncertainty,” he explains.“ If the culture of the organisation is to punish every mistake as a failure, rather than asking whether the uncertainty was properly understood, whether the decision was reasonable given the information available and what can be learned from the outcome, then leadership is actively working against the organisation’ s ability to use predictive intelligence well.” But broader culture is imperative too.
“ The right culture is one where exploration is part of the process and where post-analysis is focused on improving the way uncertainty is understood and managed,” says Juan José, who is not about to let the line of questioning coast on autopilot either.
“ I would challenge the idea that the objective is simply to‘ get out of the pilot stage’,” he highlights.“ In this age of fast code, agents and rapidly evolving
AI capabilities, organisations actually need to become comfortable living in something that looks a little like a permanent pilot stage. Products, models and ways of working are going to keep changing. If you are waiting for everything to become perfectly stabilised before doing a huge rollout, you are probably moving too slowly.”
Juan José states the distinction is between a pilot that matters and a pilot that does not:“ A useful pilot introduces something into the organisation so you
70 October 2026