How to train your team to use AI

• In short: useful training starts from real roles (exec, ops, technical), not from a generic ChatGPT course for everyone.
• Tie training to an adoption roadmap: people learn best on cases from their own work, not in the abstract.
• AI literacy is no longer optional: the EU AI Act introduces a staff-training obligation that already applies.
• Measure the outcome by real usage and cases solved, not by how many people "attended a training".

Why is a generic AI course for the whole team not enough?

A single course, delivered the same way to everyone, fails because a director, an operations person, and a developer have completely different questions about AI. The director wants to know what to adopt and where the risk is; the operations person wants to get concrete tasks done faster; the developer wants to integrate AI into the product correctly and safely. A course "about ChatGPT" gives everyone the same level of generality, which is to say nothing useful for any of them. Demand for AI training has risen sharply precisely because companies feel the pressure to adopt, but many buy catalogue courses and are surprised that nothing changes in day-to-day work. Training that works starts from "who makes what decision with AI in their role" and builds around it.

What does a role-based AI-literacy plan look like?

Split the team into three groups, each with its own learning objective and a concrete result it must produce. The table below is the structure we use as a starting point; we adjust it to each company's reality.

The key difference from a catalogue course: each group leaves the training with something usable, not just a certificate of attendance. For non-technical staff, always add the verification part — AI gets things wrong with confidence, and people need to know when and how to check what it produces.

What AI-training obligation does the EU AI Act impose?

The EU AI Act introduces an AI-literacy obligation: organisations that use AI systems must ensure that the staff involved have a sufficient level of understanding of the technology and its risks. This obligation already applies, before the rest of the regulation becomes fully applicable in August 2026. In practice, training your team is no longer only a productivity investment but also a compliance requirement. Document what training you provided, to whom, and when — just as you document any other compliance measure. For the full preparation context, also keep the EU AI Act compliance steps in mind.

How do I measure whether the AI training worked?

The number of training attendees says nothing about the result. Measure real usage a few weeks later: how many people actually use AI in their work, on which tasks, how much net time they save, and how many cases they resolve without escalating. For the technical group, the measure is whether a prototype or a real integration appeared. For management, whether an adoption roadmap with priorities and an allocated budget exists. Good training leaves visible changes in the workflow behind it; weak training leaves only certificates. Set the success threshold in advance, exactly as you would for a pilot.

What is the next step?

If you want your team to use AI for real, not just tick a course off a list, start from roles and from their concrete work cases. At Sapio we run workshops and training sessions tied to an adoption roadmap and to the AI-literacy obligation — see our AI services. Then book a free discovery call, where we work out which roles you have, what level they need, and what concrete results we are after. The initial call is free.