Julia Barham's guidance on where AI belongs in discovery workflows: it's legitimate to use AI to help process and synthesize research data, but the synthesis conversation — what the data means, whether it changes strategy — should still happen with the team, not be handed to the model alone. 'AI is there to assist you... you can use it for synthesis, just don't use it to do everything in synthesis — use it with the team, synthesize with the team.' The risk isn't using AI on data, it's letting AI's synthesis stand in for the team's own sensemaking, which is closely related to Caution Against Abdicating Product Sense to AI Tools and to Randy's complementary check, Calling BS on AI Output (Critical-Thinking Check).