Training Your Team to Work with Agents
Design training that changes how people work: hands-on sessions, shared prompt libraries, review norms, and ongoing practice.
The usual approach to AI training is a one-hour session: someone shares their screen, shows a few impressive prompts, takes questions, and sends the slides afterward. Attendees leave impressed. A month later, almost nothing about how they work has changed, because watching someone use an agent teaches you about as much as watching someone swim.
Training that sticks looks different. It's built on people's own work, it's hands-on from the first minutes, it gives people shared materials they keep using afterward, and it continues as a habit rather than ending as an event. This lesson shows you how to design that kind of training: the session format, a curriculum by skill level, a shared prompt library, review norms, and the ongoing practice that turns a session into a skill.
Why most AI training doesn't stick
It's worth naming the failure modes, because a well-intentioned session can hit all of them:
- Demos instead of practice. People watch, nod, and never try it themselves under realistic conditions.
- Generic examples. "Write a poem about our product" doesn't connect to anyone's Tuesday afternoon. People can't transfer an example they don't recognize.
- One level for everyone. The beginner is lost and the power user is bored, so both tune out.
- No follow-up. Without a next step, new habits lose to old ones within days.
- Teaching tools, not judgment. People learn where the buttons are but not how to tell good output from plausible output, which is the skill that matters most.
The fix for every one of these is the same principle: train on real work, with real practice, and keep going.