An AI assistant may know that a customer has visited the pricing page, struggled with a task, or recently contacted support. That context could help resolve a problem without making the customer repeat themselves—or lead to an unnecessary upgrade recommendation. The difference depends on what information the assistant receives, how it interprets it, and whether the customer can correct its assumptions. As AI assistants become more capable of remembering preferences and taking actions, customer data can shape far more than recommendations. Product leaders must decide what context is useful, what should be remembered or forgotten, and which conclusions an AI should be allowed to draw. This panel will explore how companies can use customer data to improve AI experiences while maintaining accuracy and customer control. When should past behavior influence a response? What should an assistant remember, ask again, or forget? How can teams measure whether additional context actually improves outcomes? And who controls the customer profile and AI-generated memories built from it? Key takeaways - Distinguishing observed behavior, inferred preferences, and explicit instructions. - Deciding what AI should remember, refresh, correct, or forget. - Testing whether customer context improves outcomes. - Understanding customer control over data, interpretations, and AI memory.