Agent Experience: The Next Layer of Product Design (Wednesday session)

Designing products with AI agents means designing how people delegate decisions and stay in control. An agent can complete a booking correctly while committing someone to a non-refundable rate before their plans are confirmed. The action may pass every system permission check and still go beyond what the person intended to delegate.

For UX and product designers, researchers, design leads and product managers, this talk explores how research, interaction design and evaluation can address that gap. It focuses on making users’ intentions and boundaries explicit in product decisions, so teams can evaluate success beyond task completion.

Attendees will learn to:

1. Use UX research to uncover what people are comfortable delegating and translate those findings into clear product boundaries.
2. Apply a framework based on consequences, reversibility and clarity of intent to decide what a product may do independently, when it should seek confirmation and how users can intervene or recover.
3. Evaluate whether an agent’s decisions respected the user’s intent and constraints, turning those expectations into criteria a product team can test.