Why most AI training doesn't work
Tools change every week. Principles don't. Hands-on or nothing.
Most AI training teaches buttons. Where to click, which feature to switch on, what the menu looks like. In a month it's useless, because the tool looks different.
I've taught courses for the public, for the financial sector, for a government office, and for a construction alliance. The pattern is always the same. What people memorize doesn't work. What they get their hands on and what survives the next update does.
The tool I show today looks different tomorrow
On one course a tool updated overnight and half the features had moved. If the whole course had rested on that menu, I would have been sunk. It didn't. It rested on the principle underneath the menu, and that didn't change.
So I don't teach a catalog of features. Features change every week. I teach a way of thinking that stays even when the interface gets redrawn.
Slides never taught anyone to work
A passive course is an afternoon of entertainment and zero change on Monday. Nobody learned to ride a bike from a presentation about balance.
My courses run hands-on. Everyone at their own computer, on their own tasks. Not a model example from another field, but that person's real work. When they do it themselves and it works, it sticks. When they only see it on a screen, it fades within a week.
A prompt framework that survives every update
One thing I teach in every course, because it holds across all tools: ROLE + CONTEXT + TASK + FORMAT + CRITERIA.
Give AI a role to speak from. Give it the context of the situation. State the exact task. Set the output format. And add the quality criteria it should hold itself to. This framework works in ChatGPT, in Claude, in Gemini, and in whatever comes next year. The menu changes, this doesn't.
A bad prompt says "write an article". A good prompt has a role, context, task, format, and criteria. That difference is the whole field.
A tailored course, not general theory
Before a corporate course I always do a pre-call. I map their systems, processes, and real use cases. The course then deals with their work, not a general AI lecture they could find on YouTube for free.
Participants leave with their own prompt pack for their own tasks and with a roadmap of what to roll out now and what in a month. Not with a feeling that it was interesting. With things they'll use tomorrow.
How to spot good training
Ask three things. Will I get my hands on my own task during the course? Will I learn a principle, or just where to click? Will I take away something I'll use on Monday?
When the answer is yes three times, it's a good course. When it isn't, it's a paid afternoon after which nothing changes. Tools change. Principles and hands-on don't.