Building trust in AI isn't about transparency; it’s about communication design. Whether it’s an autonomous vehicle or a generative agent, the challenge is the same: how much should a system communicate? Drawing from years at Waymo, this session explores a framework for trust. We’ll move beyond automation toward legibility, where users understand intent rather than just monitoring tasks. Learn to calibrate feedback to avoid alert fatigue and build AI products that feel predictable and fundamentally trustworthy.