How can employees participate in an AI project before the tool is already built?
Choose one task with the employee and ask them to show it from the incoming request to the completed result. Let them explain which parts are difficult before you propose where AI belongs.
Liz Bacelar described changing her approach from delivering an AI solution to building it with the people doing the work, and her example connected learning with the employee's actual process. That's a different starting point from a general demonstration followed by a request to adopt a tool.
During the discussion, demonstrate one small possible use, like organizing supplied information or preparing a draft for review. It helps to explain what it receives, what it produces and what it can't establish. The employee can then compare the result with what the job requires.
Their corrections are design work
Ask them to point out an error or exception the demonstration missed. That correction is part of the design and part of the learning, so explain how it changes the proposed workflow, and don't just say the model needs improvement.
Keep a record of the choices the employee helped make. Which step should remain theirs, and what information should stay out of the tool? Also note what would make the proposed change unhelpful.
A working demonstration can make AI less abstract. Participation matters because the employee can still influence the design, and watching someone else finish it doesn't give them that.
What to try this week
Ask one colleague to show you a task from start to finish, and write down the steps they'd want to keep for themselves.