Most AI rollouts die in one of two ditches: a free-for-all with no guardrails, or a policy nobody reads. I build the middle path. A practice that gets people using AI on real work, safely, and proves it was worth it. Here is the method.
I begin with what people actually do all day and where it stalls. AI earns a place only where it removes a real bottleneck. No tool in search of a problem.
A short, readable acceptable-use policy, a clear privacy line, and a yes / no decision aid. People move faster when the boundaries are obvious.
Adoption is a behavior problem, not a memo. I build prompt libraries, templates, and hands-on training so the right move is the easy move.
Track real adoption and real outcomes, not seat licenses. Keep what works, cut what does not, and say so plainly.
Concrete artifacts a team keeps using after I leave:
Assess to scale, with owners and milestones.
One page. Acceptable use, privacy, the yes / no line.
Role-based prompts and templates staff actually run.
Workshops, job aids, and the product UX to match.
I built this practice from the ground up for a national organization and its network of more than 125 partner organizations across 30 states: the strategy, the governance, the staff training, and the UX across its platforms, LMS, and portals. I also use AI in my own work every day, for storyboards, prototypes, and pressure-testing a plan before a human ever sees it. It earns its place the same way any tool does. It ships work faster without making it worse.