Event storming was always about surfacing the questions nobody wrote down, so it’s a good fit for AI to accelerate. The risk I’d watch is the model generating plausible events that smooth over the real disagreements in the room. The friction between domain experts is the signal, not noise to optimize away.
Event storming was always about surfacing the questions nobody wrote down, so it’s a good fit for AI to accelerate. The risk I’d watch is the model generating plausible events that smooth over the real disagreements in the room. The friction between domain experts is the signal, not noise to optimize away.
That's a great point.
I think the distinction becomes much more important when the side effect has real-world consequences rather than just data consequences.
An email can be retried. A report can be regenerated.
A payment, a customer call, or an autonomous agent taking action is different because the cost of being wrong is much higher.
At that point, the question is no longer just sync vs async. It's about approval boundaries, reversibility, and risk ownership.