We have given every employee a virtual assistant. We have integrated it into the Office suite, the browser and the ticketing system. Perhaps we even use it to book lunch at the staff canteen.
AI has become part of our working tools. Have we also invested in training the people who use them?
Assigning a licence is a starting point. Learning to use it takes an equally practical investment.
Beyond questions and answers
If we use an assistant only as a search engine, we risk exploring just a small part of what it can do.
We can ask it to help prepare a meeting, compare documents, turn notes into a procedure or identify missing information in a problem description.
To do this well, we need to learn how to describe the work: the result we want, the data we have, the constraints we must respect and how we will check the response.
These are basic prompting techniques. Their value becomes clear when we apply them to everyday tasks.
An hour a month is enough to start
We can begin by dedicating an hour a month to training. A short, regular session built around the participants’ work. A starting point to adapt to the group’s needs.
A traditional classroom is not essential. A video call, a shared screen and a real task can be enough.
Take a generic request:
“Write a reply to this ticket.”
Let’s make it more useful:
Draft a reply using the information in the ticket
and the attached procedure.
Separate checks already completed from the next steps.
If essential details are missing, tell me what to ask for.
Use clear language and do not promise resolution times
that have not been confirmed.
We compare the responses, discuss the errors and check the result. This is where training becomes working practice.
Learning together
Working through a real task, we learn to give instructions, recognise errors and check responses.
Autonomy at work
AI drafts, compares and suggests. The person assesses, corrects and decides.
Results to assess
Quality of work, corrections needed and time saved.
Time saved
New goals, better processes and more room for people.
Can AI train us on its own?
On its own, no.
AI can suggest exercises, explain a technique and help us experiment. But putting it entirely in charge of training also means asking it to judge whether its own responses are adequate.
We need discussion with someone who understands the process, can recognise an error and can explain why a convincing response may be unsuitable for the job.
Self-directed learning can support these sessions. The opportunity to ask questions, watch a colleague work and discuss a real case remains essential.
Value results, not time
Learning to use assistants more effectively can reduce repetitive work and make room for tasks we keep postponing: improving a procedure, investigating a problem or developing a new idea.
Or, why not, having more free time.
Measure the benefits in actual work: the quality of the result, the corrections needed and the total time taken to reach a usable outcome. Avoid promising the same productivity multiplier for everyone.
Put AI at the service of our colleagues.
And start assessing results rather than time spent at a desk.