A demonstration is only the beginning
Watching someone else use a tool is different from using it on your own work. An employee may leave a session knowing what AI can do, yet still be unsure what information they may enter, how to check the answer, or when a manager needs to approve it.
Start with one familiar task
Choose a task the employee already understands: preparing a customer update, turning approved notes into a draft, or checking a routine report. Agree what a good result looks like before opening the tool. This gives the employee something concrete to judge.
Practise the whole job
The exercise should include more than generating an answer. Find the source information, choose what may be used, produce the draft, check the facts, and make the final human decision. A fluent answer with an incorrect date is still incorrect work.
Ask the employee to show you
After guided practice, ask the employee to repeat a similar task and explain the checks they made. Record what they can do and where they still need help. A course certificate should reflect the course requirements, rather than attendance alone.
Plan the first use after training
Choose a normal workday when the employee will use the skill again. Make sure the approved account is available and the manager knows what support to give. One repeatable task is a clearer starting point than asking everyone to use AI more.
What informed this approach
Docebo's 2026 research reports a gap between training and employees' ability to apply AI in their role. That finding motivates job-specific practice; it does not establish that any one provider's course guarantees a better result. Read the publisher's research summary.