Why adoption is the real deliverable

A pilot that measures well and then goes unused has delivered nothing. The value of AI in a process shows up only when the people running that process change how they work: they check the draft instead of writing it, they ask the system before they ask a colleague, they flag what it gets wrong. That is behaviour, and behaviour is trained, not installed.

The resistance is rarely ideological. It is practical. People do not know what the tool is for, what it may not be used for, when to trust it, and who to tell when it is wrong. Answer those four questions for every role and adoption follows.

What the EU AI Act asks of you

Since 2 February 2025, Article 4 of the EU AI Act requires providers and deployers of AI systems to ensure a sufficient level of AI literacy among the staff who operate them, taking into account their role, their technical knowledge and the context the system is used in. It applies to any company using AI at work, not only to those building it.

The Act does not prescribe a course. It expects that you can show your people understand the systems they use well enough to use them responsibly. In practice that means training matched to roles, and a record that it happened.

How we train a team

Training is the third step of our approach, and it runs alongside the pilot rather than after it.

Workshops by role. Sales, marketing, operations and leadership do not need the same session. Each group works on its own process, with its own data, on the tools it will actually use. A two-hour session on a real task beats a full day of slides.

Internal champions. In every team we identify one or two people who pick it up fastest and give them more: deeper sessions, early access to changes, and a direct line to us. They answer the everyday questions long after the project ends.

Rules people can remember. What the system may see, what it may not, and what must be checked by a person before it leaves the building. Written on one page, agreed with the team, and revisited when the tools change.

Documented. Who was trained, on what, when, and what they can do now. This is the record the AI Act expects, and it is also how you know where the gaps are.

What that looks like in practice

• A role-by-role list of what each team will use AI for, and what it will not.
• A short usage policy in plain language, signed off by the people it applies to.
• Two champions per department, named and given time for the role.
• Training sessions run on the team’s real data and real tasks, not demo material.
• A literacy record per person, kept current as the tools change.

What you can skip

You do not need a company-wide AI course, a certification programme or a change management office. You need the people who touch the pilot to know what it is for, what it must not do, and who to call. Start there and widen it as the next process comes on.

The bottom line

The model will be replaced. The integration will be rebuilt. What stays is a team that knows how to work with AI and is allowed to. Train for that from the first pilot and every one after it starts with people who already know what they are doing.

If you want to know where your team stands, the readiness assessment includes a literacy check by role.