Your team's AI adoption has one lever: the manager

If you want to know whether a team will actually adopt AI, don't look at the tools or the mandate. Look at the manager. Gallup's 2025 research isolates the variable cleanly: employees who strongly agree their manager supports the team's AI use are 2.1 times more likely to use AI a few times a week or more, and 8.8 times more likely to strongly agree AI helps them do their best work. That's not a nudge; a multiplier that size separates a tool people reach for from one they ignore. And here's the opening: only about 28% of employees say their manager actively supports their team's AI use. The lever that moves AI adoption more than anything else is sitting there, mostly unpulled, on most teams.

This isn't an argument for more dashboards or a sterner email from leadership. It's an argument that AI adoption is decided one manager at a time, by what that manager visibly does.

AI adoption is still mid-curve: Gallup found only about 23% of US employees use AI at work a few times a week and 10% daily, so there's real headroom to move. The two instincts companies reach for to move it, a top-down mandate and a usage dashboard, both backfire, because forced tools feel like more work and being watched suppresses the behavior you're measuring. The manager actually moves adoption because a manager doesn't mandate or monitor; they support. They make a new tool feel safe to try, point it at a real task, and notice when it helps. None of that shows up in a policy document, and that absence explains why the policy-and-dashboard approach keeps disappointing while the manager keeps working. This is the human core of any AI transformation, and it's why adoption lives or dies in the middle of the org.

There's a fair pushback worth meeting head-on. Adoption does have a real grassroots side, employees often experiment with AI well before any formal program, and leaders consistently underestimate how much their people already use it. And managers are genuinely overloaded; research describes a donut hole where the C-suite invests and younger workers are fluent, but the middle managers who must orchestrate the change are the ones stalling, under real constraints and without enough time or support. Both points are true, and neither sinks the case. Grassroots energy stalls at a ceiling without a manager to give it air cover, protected time, and reinforcement, which is exactly what the 28% gap measures. And the burden objection is why the actions below are deliberately lightweight. The manager's job here isn't a new initiative to run; it's a lever, and a lever multiplies a small input. You're removing friction and pointing at one task; no program required.

The line that separates a manager who moves AI adoption from one who doesn't is coach versus cop. The cop tracks who's using which tool and frames AI as something to audit, which erodes the psychological safety people need to try an unfamiliar tool and admit when it didn't work. The coach does the opposite: models curiosity, asks what the team tried, helps whoever is stuck, and amplifies a real win. Microsoft's research found that when managers actively model AI use, their teams report more psychological safety and faster adoption, and a named champion on the team accelerates it further by showing peers AI on their own tasks. The whole posture is enablement, it levels people up without ever watching them, and it's what makes AI for teams stick across every role. Start Monday with the smallest version: pick one task, name it, protect thirty minutes, and ask about it on Friday.