Case Study: EPFL, administrative department
Thirty staff trained on their own tasks, with a prompt library ready before the session
Situation
The École polytechnique fédérale de Lausanne wanted to train 30 staff in its administrative department on generative AI. Such a department handles official letters, case files, internal procedures and personal data: confidentiality is a daily constraint. And thirty people trained together means thirty individual practices already coexisting.
Objective
Replace scattered practices with common ground: a plain-language understanding of how generative AI algorithms work, a clear view of their strengths and weaknesses, consistent usage across the department, respect for ethics and data confidentiality, command of prompt writing, and a workshop on cases specific to administrative staff.
Staff trained
30
Deliverable handed over
A prompt library
Needs gathering
Beforehand, by form
Use case sign-off
By the head of department
Process
Gather the use cases and have them signed off
A form sent to the 30 staff beforehand captured their real tasks and the irritants they wanted to tackle. The cases selected for the workshop were then approved by the head of department, so that nobody worked in the room on a task management would not have endorsed.
Translate the department tasks into AI use
Rather than starting from a generic list of administrative use cases, the real tasks of the department were understood and then translated one by one into concrete uses, to build training and a workshop that could be applied straight after the session.
Hand over a ready-to-use prompt library
A prompt library specific to the department was built beforehand, so that the teams had a usable starting point by the end of the training instead of everyone rebuilding their own prompts in the following weeks.
Results
A needs-gathering form sent to the 30 staff before the training, and workshop use cases approved by the head of department
A programme covering how the algorithms work in plain language, their strengths and weaknesses, consistent usage, ethics and data confidentiality, and prompt writing
A workshop built on use cases specific to administrative staff rather than on a generic list
A prompt library handed over to the department, usable from the end of the training instead of rebuilt afterwards by each person
Contact us
- Want to upskill your teams?
Let's schedule a meeting to assess your training needs
- Want to identify high-ROI use cases for generative AI?
Let's schedule a strategic call