WHAT YOUR PEOPLE ACTUALLY USE
You are already paying for tools that most of your people open once and never return to. You leave with those tools in daily use where the work actually happens: measured before, designed into the workflow, and measured after. The gain comes out of spend you have already committed.
THE THESIS
Production is the halfway point. A system can be integrated, compliant and running, and still change nothing, because the people it was built for keep working the way they always did. Usage is not a communications outcome. It is a design outcome.
When we measure organisations, the barrier is rarely willingness. What we find instead are gaps the leadership team could not see from where it sits: competence lower than assumed, real uncertainty about what data may be shared with which tools, and high motivation with no plan to act on it.
None of that is fixed by a launch email. Very little of it is fixed by a course alone.
FIVE STEPS, IN ORDER
Adoption starts as a baseline, not a campaign. Before anything is designed we establish where the friction actually sits, who is already using AI quietly, and what is stopping the rest. The answers come from your own organisation, not from a template.
AI that lives in a separate window is AI that gets forgotten. We put it inside the systems and steps people already use, and we settle the unglamorous questions that decide adoption: who does what differently on Monday, and who owns the new step. A tool that lives outside the workflow dies outside it.
The fastest way to freeze usage is uncertainty about what is allowed. People who don't know which data may go where choose the safe option: nothing. A workable policy and plain answers unlock more usage than any campaign. Uncertainty freezes use faster than resistance does.
Most mandatory e-learning is clicked through, not completed. We build the opposite: leadership sessions delivered by Ampliro in the management team, and role-specific programmes for the wider organisation through AIUC, our education arm. Applied to your workflows, documented, and measured for effect, not attendance.
Adoption is a number, not a feeling: usage in the actual workflow step, time and quality in the task, confidence before and after. The follow-up points are defined in the engagement, so you know whether it stuck, and why.
WHERE WE PLUG IN
Our own AI implementation or one already underway: adoption designed in from the start, not bolted on at launch.
You launched, they didn't come. We measure why, fix the workflow and the clarity, and re-measure. Sometimes the honest answer is that the tool was wrong. You will hear that too.
The baseline and the gaps, so the rollout plan is built on how your organisation actually works, not on how the vendor assumed it would.
QUESTIONS
AI change management is the work of making a tool actually used, as distinct from making it available. A rollout ends when the system is live; change management ends when the work has changed. In practice that means a baseline of where the friction sits, the tool designed into the systems people already work in, a plain answer on what data may go where, and a named line manager who owns the new step. Then the same measurement again, to see whether it held.
AI tools already in production go unused when the workflow, the clarity or the competence is missing, and AI adoption is fixed in that order. We measure where it stalls, in your own workflows and with the people who work in them. Three causes recur: the tool sits outside the systems people already work in, nobody knows which data may be used, and no line manager owns the new step. Where personal data is involved, the Swedish authority Integritetsskyddsmyndigheten supervises it, and a plain answer on that unlocks more use than any campaign.
AI adoption is measured against numbers agreed before the work starts: the share of cases where the actual workflow step is done with AI, time and quality in the task, and people's own confidence before and after. There is a baseline first and a measurement afterwards, at follow-up points set in the engagement. The same evidence feeds a management system to ISO/IEC 42001 if you run one, although that standard is not harmonised under the EU AI Act. No vanity metrics.
The AI literacy duty in Article 4 of the EU AI Act asks you to take measures that promote AI literacy among your staff and among others who work with AI on your behalf, and it has applied since 2 February 2025. Since July 2026 the wording is explicit that no particular level has to be guaranteed in any individual. In practice it still means two things: the leadership team understands what it has decided, and every role that touches an AI system knows what applies to that role. It is one of the strongest reasons to build competence and adoption together.
AI adoption work needs three kinds of people in the room: a sponsor in the leadership team, the line managers who will own the new way of working, and the people who do the work, because that is where the insight sits. Public bodies, and deployers running creditworthiness assessment of individuals, also owe a fundamental rights impact assessment under Article 27 of the EU AI Act from 2 December 2027, and that assessment needs the people who know the workflow.
AI adoption work is a defined engagement at a fixed price, not an ongoing programme. The weight sits around deployment, with follow-up points that are agreed and finite, and the engagement ends on purpose. Where a date applies to you, such as the transparency obligations in Article 50 of the EU AI Act from 2 August 2026, the follow-up is placed before it. If a later initiative needs the same work, that is a new engagement, not a subscription.
AI training alone rarely gets adoption going, because training is one lever among several and usually not the first. The gaps we measure are normally workflow, clarity and ownership, and no course fixes those. Where training belongs in the answer, Ampliro delivers the leadership sessions and AIUC delivers the role-specific programmes, measured for effect rather than attendance. Which capability each role actually needs is settled in the AI competence work.
INSIGHTS
More than half of employees say they conceal their AI use. That turns every figure built on asking into a floor rather than a level, and it is the whole difference between a diagnosis and a score.
Read the analysis
In Sweden 88 percent say employees should be involved when new technology is introduced. A Swedish union has measured what actually happens, and the distance between the two is not a matter of opinion.
Read the analysis