UXDX Community:Actionable Design, AI-Native Products

Juhan Sonin will show how design actually gets things done inside healthcare and public systems—where nothing is clean, politics are thick, and progress comes from pushing through bullshit, not waiting for permission. The key isn’t just starting, it’s starting with an agenda: a point of view, a spine, and the integrity to hold the line when the system tries to grind you down. Through real projects in public health, civic tech, and AI, Juhan shares how scrappy, messy ideas turned into shipped services for millions, often spawning better projects than the ones we started with.

AI-native products behave differently from traditional software. They adapt, vary, and respond to context, requiring teams to rethink how they design, develop, and evaluate quality. Linear flows break, predictability erodes, and quality must be evaluated across system behavior over time rather than single interactions.

The SHIP Interaction Intelligence Framework, developed through Bianca Work’s work at Indeed, Splunk, and Alvio, provides a practical structure for shipping and scaling AI-native product ecosystems. SHIP helps cross-functional teams assess four critical areas: system behavior, human expectations, interaction design, and product instrumentation as products evolve.

This session examines how non-deterministic behavior appears in real products, where confidence breaks down, and how failure modes and drift should be designed for and monitored. Attendees will leave with a shared, actionable understanding for designing, developing, and evaluating AI experiences and shipping systems that remain trustworthy, reliable, and resilient over time.