The playbook for spreading AI inside a company keeps producing the same failure.
The usual approach is to find one enthusiastic employee, protect 20 percent of their time for AI experimentation, and wait for the habit to spread to the rest of the team. One person in a recent discussion of this approach said they had already run that play, during the robotic process automation boom a few years back. Leadership had agreed to protect a fifth of the champion's schedule for RPA work. A month later, the champion's manager quietly took that time back.
The pattern keeps repeating because protected time is never actually protected once it collides with a manager's own targets. The diffusion model also has a built-in accountability gap: if adoption is everyone's job, it ends up being no one's job, and the stall goes unnoticed until the pilot quietly dies. Without a budget line or a named owner, a champion's extra hours are the easiest thing on the org chart to reclaim.
It's the same resourcing trick companies ran with RPA a decade ago, and it produced the same result: enthusiasm in a slide deck, then a quiet reversion to how things were always done.