When everyone has AI and the company still learns nothing

Organizations are entering the 'messy middle' of AI adoption where tool proliferation (Copilot, Claude, etc.) occurs across teams at uneven rates, yet fails to create measurable organizational learning or strategic advantage. Traditional change management approaches—communities of practice, enablement programs, dashboards—are too slow to capture the real learning that happens in individual workflows and code reviews, making it critical for IT leaders to shift from tracking tool usage metrics to understanding how teams actually integrate AI into their decision-making loops. This misalignment means companies risk spending millions on AI licenses while failing to systematize the tacit knowledge and friction-driven insights that create genuine competitive advantage.