- The context
- An industrial SME in metalworking wanted to launch its data and AI transformation without falling into the disconnected "AI project" trap. The stance: start from the teams' real pain points, not from a tool in search of a use.
- The lived pain
- The ideas and problems already existed, but scattered: information too slow to find, re-keying, underused data, inconsistent quality checks, an underused history of quotes, know-how hard to pass on.
- The approach
- Time on the floor to understand the real work, an AI literacy step with no magical talk, then workshops where the teams themselves surface their pain points. The raw ideas are grouped into use cases, scored on two axes (impact and feasibility), then turned into a roadmap.
- The resultmeasured
- 151 ideas surfaced by the teams, 14 contributors, 34 structured and prioritised use cases. Concrete material, not a list of intentions.
- What it changes for the team
- The teams don't undergo the transformation, they feed it. AI doesn't come down onto the floor: it starts from the floor. The result: more buy-in, realistic use cases, and far less distrust of the tool.
- The limit, said plainly
- A workshop isn't enough. Without someone keeping it alive afterwards, even the best-scored ideas stay neatly arranged sticky notes. The value comes from the ritual that turns them into decisions, then results.