You bought the licenses. You sent the announcement email. Three months later, usage data tells the real story: four people out of forty are actually using the tools. Sound familiar?
Why this happens. Most AI rollouts fail because they're deployed as technology projects when they should be behavior change projects. Your team doesn't need to understand what a large language model is. They need to see how it saves them 30 minutes on the thing they hate doing every Tuesday afternoon.
The fix is specific, not general. "We have AI tools now, go explore!" doesn't work. What works: identifying three to five specific, recurring tasks per role and building explicit workflows. Not "use AI for writing," but "here's how to draft the weekly client status email using this exact prompt template."
Champions, not mandates. Find the two or three naturally curious people. Give them extra support. Let them develop workflows. Then have them show their peers. Peer demonstration is ten times more effective than top-down mandates.
Measure the right thing. Don't measure "how many people logged in." Measure "how many hours did the team save this week." When people see real time savings tracked publicly, the holdouts start paying attention.