Mansi: The biggest surprise has been how often the most valuable insight in the room comes from the participants, not the technology. During a generative AI session with a mid-sized insurance firm, a senior underwriter stopped an exercise halfway through. She looked at an AI-generated claims summary and said, ‘This would fail our subrogation review — the model doesn’t know that step exists.’ The room went quiet. Then the rest of the underwriters started jumping in: one flagged an exception the model had missed, another pointed out a compliance issue, someone else described a customer scenario that would change the whole outcome. In under ten minutes, we learned more about that process than weeks of documentation had told us. Nobody in that room was resisting AI. They were doing exactly what experienced professionals do — applying judgment. That’s the moment I look for in every session: people arrive assuming I’m there to teach them about AI, and by the middle of the session, the learning is going both ways.