From awareness to action: training Kemppi teams to use Copilot in real work

At Kemppi, the goal was clear: move beyond general AI awareness and help people actually use Copilot in their daily work.

Reaktor Ecosystem company Splended designed and delivered training in both Finnish and English across teams with very different starting points. Some participants were already experimenting with AI tools, while others were taking their first steps. This mix is typical in many organisations right now and it shapes what effective training needs to look like.

From the beginning, their coach Santeri Kallio focused on practicality. Not what AI could do in theory, but how it fits into real tasks: writing, summarising, searching, structuring work, and increasingly, building agents that can take action. As one participant put it, the training helped them “start using Copilot for more than just basic chatting.” Another described the agent part as “a clear level-up.”

What worked well was keeping things grounded. Participants appreciated clear explanations, concrete examples, and a calm, structured way of tutoring. Many highlighted that even if they had followed AI development already, they still learned something new. For those newer to AI, the sessions worked as a solid starting point: “a very good starter package” that made the topic more approachable.

People soon started to connect AI to their own work. In one exercise, participants used Copilot to design workshop agendas for their own teams, including slide decks they could run with minimal editing. Others started mapping where agents could fit into their daily work: meeting minutes, recurring reports, searches across scattered information.

At the same time, the feedback highlighted something important. When people are at different levels, one format does not fit all. Some participants found the beginning too basic, while others needed exactly that foundation.There was a clear request to split sessions between beginners and more advanced users and to go deeper into practical use cases for those already using AI.

Another clear signal was the interest in continuing. Several participants asked for follow-ups after a few months, once they have had time to test things in their own work. This reflects a broader pattern we see: real learning happens when people try things, come back with questions, and build on their own experience.

Participants also took AI’s limitations seriously. One exercise made the point stick: Copilot invented a completely wrong answer to a simple factual question, then got it right the moment web search was enabled. Misinformation and hallucinations stayed top of mind throughout, and people were already treating AI more as a sparring partner than an authority.

For us, this collaboration with Kemppi reinforces a simple point: AI training is no longer about introducing the topic. It is about helping people move from curiosity to capability. That means focusing on real tasks, supporting different starting levels, and creating space for follow-up learning.

The next step is not more information. It is helping teams build habits around using these tools and gradually moving from individual use to shared ways of working.