Large Language Models like ChatGPT, Gemini, or Claude are very good at generating events, actors, and processes on demand. Big Picture EventStorming is very good at exposing what people really know, what they only assume, and where the real problems live. In this workshop, we put the two together and ask a simple question: Can AI actually help, or does it just create smarter looking slob?

In this hands-on lab we will simulate a Big Picture EventStorming around a real world use case. We work under a realistic constraint. You do not have a dedicated domain expert in the room. Instead, you only get a short written brief or notes from a rushed interview with the expert, and the LLM helps fill in the blanks. Instead of starting from a blank wall, each group is supported by an LLM that can suggest domain events, actors, systems, and problems on request.

Together we will:
Use AI to bootstrap the chaotic exploration phase and then clean up hallucinations and weak assumptions.
Enforce a shared timeline and highlight pivotal domain events, this time with every AI suggestion clearly marked as strong, inferred, or speculative.
Map people and systems, paying attention to sociotechnical aspects like ownership, handovers, and conflicting goals.
Surface hotspots, bottlenecks, and opportunities, and then reflect together on how useful the AI aided discovery really was and how far they would trust the resulting picture.
Throughout the workshop we will treat the LLM as a junior domain expert that is allowed to be wrong but must be transparent about its assumptions. We will close with a joint reflection: When is AI a useful accelerator for Big Picture EventStorming, when is it risky when you only have a brief use case description or a single short expert interview, and how could you safely experiment with this format in your own organization or classroom.