AI tools don’t create value. People do.

AI adoption is on every leadership team’s agenda. Organisations have invested in tools, launched pilots, and rolled out programmes. Yet when you look at how work actually gets done, the change often remains marginal.

The reason is rarely the technology itself.

According to Deloitte’s State of AI in the Enterprise 2026 report,, the AI skills gap is the biggest barrier to AI integration, ahead of challenges related to technology, data, and governance.

The challenge is not the amount of training but its impact. Too often, training is treated as a standalone initiative rather than something that should change how work is done, how roles are defined, and how processes are designed. Without that connection, skills don’t translate into action.

Completing a course is not the same as building capability

Most AI programmes still follow a traditional model: select content, deliver training, measure completion. The same content is offered to everyone, regardless of role, experience, or actual needs.

The issue is rarely the quality of the content. The issue is that generic training does little to help people apply what they’ve learned in their day-to-day work.

AI is not learned in a classroom. It is learned by using it in real tasks, in real situations.

Experimentation is an important first step — it helps people understand what AI can do. But experimentation alone doesn’t build capability. It only answers the questions we already know how to ask.

Real capability develops through repeated practice, feedback, and guided application. That is when AI starts to change how work gets done.

What does this mean in practice?

If your AI investments are not delivering expected results, the answer is unlikely to be a better tool or a larger budget. It comes down to how capability is developed and whether it is developed systematically at all.

Three shifts matter most.

From generic to personalised. Learning should be tailored to an individual’s current skill level, role, and objectives. A single programme for an entire organisation is unlikely to build the capabilities needed to transform work.

From watching to doing. Capability is not built by watching videos or completing courses. It grows through repeated practice on real work tasks. Practice is not an addition to training. Practice is the training.

From completion metrics to capability metrics. Course completion rates reveal very little about actual capability. The key question is not how many people completed a course — it is what people can do differently afterward.

The organisations winning with AI are building capability, not just buying tools

Multiple studies indicate that organizations with structured AI capability development programmes, including workforce upskilling, leadership engagement, governance, and workflow redesign, are more likely to achieve successful AI adoption and realize greater business value from their AI investments than organizations that rely primarily on technology implementation.

Over the next few years, competitive advantage will not be determined by who has access to the most AI tools. It will be determined by who embeds AI most deeply into how work and business gets done.

Organisations that build AI capability systematically will do more than keep pace. They will shape how work gets done and define the standards others follow.

Written by Katja Miettinen & Stefan Heinrichs