Working with AI usually happens alone — and nobody sees how a good result came about. collab.prompting.schule turns prompting into something shared: learners tackle the same task with a language model and then compare how they went about it.
How it works
- Teachers set up a protected CollabSpace: with a task, a system prompt and a language model behind an OpenAI-compatible interface.
- Learners join with the space ID and access code and start their own CollabChats.
- The chats are assessed; afterwards learners see each other’s prompting approaches and can discuss them.
Only what is created in the CollabSpace is shared — not the private history with other AI tools. That is deliberate: in the accompanying research, around 70 % of participants were hesitant to disclose their full prompt history, yet collaborated productively in the protected space. The focus was less on the “best” prompt than on how it was revised.
The platform complements prompting.schule (external site): there, people learn prompting on their own at their own pace; here, together in class. It is published as free software and can be self-hosted by other institutions.