Absorptive capacity: the bottleneck was never the tool (corrected, verified figures)

Absorptive capacity: your bottleneck was never the choice of an AI tool.

Deloitte's 2026 survey of more than 3,200 leaders found a telling gap. Companies feel ready for AI in the boardroom but not on the ground. Strategy is the dimension they rate best prepared, yet only about a fifth say their people are genuinely ready to use the tools, and skills are now the single biggest barrier to putting AI to work.

The reflex, when a rollout under-delivers, is to blame the tool. Wrong model, wrong use case, not quite capable yet. In my experience, that is not correct. The model could do the work. The organisation simply could not absorb it.

There is an old idea in organisational research called absorptive capacity: a firm's ability to take in something new and turn it into action. It predates this wave of AI by decades, but it explains most of what stalls today.

The constraint sits not in the technology but in everything around it. The people who have to trust the output. The workflows were built for a slower, manual world. The governance team has not yet decided who is accountable when the model gets it wrong. The target operating model never made room for the new step.

You can buy capability off the shelf in an afternoon. You cannot buy capacity. It has to be built, and built slowly: through training that goes beyond the launch webinar, through redesigned processes, through clear lines on what AI is allowed to touch and who signs off on it.

This is why two firms can deploy the same tool and get entirely different results. One bolts it onto a system that cannot metabolise it. The other has done the unglamorous work of preparing the ground.

Most AI roadmaps spell out what the tools can do. Far fewer say what would have to change around them for any of it to land.

Which half is your roadmap missing?