95% of corporate AI projects deliver no measurable results (MIT NANDA). The reason is rarely the technology itself, but what happens in the month after the course: nobody is responsible for keeping the skills alive. The Champions method is the simplest and cheapest way to close that gap.
Why do the skills evaporate?
The typical sequence of events looks like this. The company runs a successful course. People go home excited, with new ways of working. In the first few days most of them try something. Then a busy week arrives, old habits take over, and after a month perhaps two or three people are using what was taught.
This is not because people are uninterested. New ways of working always lose to old habits unless something keeps them alive: fixed times, visible examples from colleagues, and someone to ask when you get stuck.
On top of that, the AI landscape changes every month. The tool that was best in January may not be in March. Unless someone keeps track, the knowledge goes out of date by itself, even while people are using it.
What is a champion, and what do they do?
A champion is an ordinary employee, not the IT manager and not an outside consultant, who receives deeper training and a clear role: to keep the department's AI use alive. In the AI Sprint we train 2 to 4 champions in each department alongside the general training, precisely so that the development continues after we leave.
The role is built on three fixed tasks that take about 2 to 4 hours a week in total:
Open clinic. frequency: weekly, 30 to 60 min · what happens: colleagues turn up with real tasks that got stuck. The champion solves them with the people, not for them
Tool review. frequency: monthly · what happens: the champion assesses what has changed in the tools, what to adopt and what to drop, and sends a short summary to the department
New-hire training. frequency: with every hire · what happens: new staff receive the department's basic training from day one instead of inheriting old ways of working
These three tasks address the three reasons skills die: the clinic keeps the habit alive, the tool review keeps the knowledge fresh, and new-hire training prevents the group from thinning out over time.
How do you choose the right champions?
The most common mistake is to pick the most technical employee. The person who is far ahead of everyone else often finds it hard to teach, and colleagues do not see themselves in them. Better criteria are these:
- People colleagues already turn to. A champion only works if people dare to ask them stupid questions.
- People who know the tasks, not just the tools. Research shows the greatest gains among a broad group, not just the tech nerds (Harvard/BCG 2023). A champion needs to understand what the department actually does.
- People who want the role. An appointed champion with no interest in it will have stopped holding clinics after three weeks.
- More than one. A single champion goes on parental leave, changes jobs or burns out. 2 to 4 in each department means the role survives staff changes.
What do champions need from management?
The Champions method costs little, but it does not cost nothing. Three things need to be clear from management, otherwise the role fades out like everything else.
First, time. 2 to 4 hours a week need to be a recognised part of the job, not something the champion does over lunch. Second, a mandate: the champion needs to be able to propose changes to processes and tools without going through three layers of approval. Third, visibility: when the clinic solves a task that saves the department an hour a week, the manager should say so at the next department meeting.
The difference is measurable. In Iceland, 80% of professionals use AI at work but only 34% have received training from their employer (Viska 2025). A company that trains and keeps the training alive with champions is not competing against perfection; it is competing against workplaces where everyone fumbles along on their own.
What you can do right now
- Find the informal champions who already exist. Ask three employees: "Who do you ask when you get stuck on an AI task?" The names that come up again and again are your starting point.
- Put one open clinic on the calendar. 45 minutes next week, open house, people turn up with real tasks. One clinic costs nothing and shows immediately whether the demand is there.
- Write down what the last course left behind. How many people use what was taught, a month or a year later? If the answer is "we do not know", that is the answer itself: nobody is responsible for what comes next.
[ Get in touch ]
Book a free assessment
90 minutes that pay off immediately: we map your AI usage, risks and 3 to 5 automatable workflows, and deliver a report within a week. No commitment.