For product designers who want AI features to be trustworthy, clear and well judged. You will learn to plan AI behaviour, make answers and intent legible, and evaluate output against real user needs.
8 lessons · 3 hr 1 min, plus 1 optional (80 min more with the optional lessons)
Start here: a quick, practical opener on judging whether AI answers are trustworthy and showing that in the interface.
Shows a process for designing AI products that users can trust, focusing on how to judge whether AI-generated answers are correct and how to make that visible in the interface. Uses Claude Design as the working tool.
Adds the HCI theory behind trust, including automation bias, over-reliance, and why more explanation does not always help.
Explores what trust means in HCI for AI-driven systems, including opacity, automation bias, over-reliance, and why more explanation does not always create more trust. It discusses design factors such as transparency, predictability, control and accountability, and the idea of appropriate rather than maximum trust.
Moves from trust to planning: choosing the type of AI a feature needs and how much agency it should have.
The video explains how to plan AI-powered apps before designing the interface, covering what type of AI a feature needs, how much agency the AI should have, and how to design human-centred AI experiences. It walks through three examples, from simple features to agentic apps, showing the reasoning behind each decision.
Applies those agency choices to communication, showing how AI systems can make their intent legible to users.
A Config 2026 talk drawing on the speaker's years at Waymo that sets out a framework for how AI systems should communicate with users. It covers moving from automation toward legibility, where users understand intent, and calibrating feedback to avoid alert fatigue.
Shifts from designing the experience to judging it, defining what good output means and evaluating against human needs.
A Config 2026 talk arguing that designers and researchers should help define what "good" means for AI systems and lead how those systems are evaluated against human needs. It is framed as an invitation for design and research practitioners rather than a hands-on tutorial.
Widens the lens to agentic experience (AX), designing for AI agents that act on human intent rather than screens alone.
A talk on the shift from designing screens and flows for users to designing for AI agents that act on human intent, which it calls Agentic Experience (AX). It is aimed at designers at different career stages, with particular attention to mid-career product designers.
Introduces the first advanced question: balancing user agency against automated magic in adaptive, learning software.
A Config 2024 talk in which the speakers present their vision of "couture software", computers that learn user preferences and adapt their behaviour and form. The talk raises how to build a design system that supports this while keeping a creative vision, and how to balance user agency and control with the magic of the system.
Optional deep dive: governance, task-completing agents and AI UX best practices, for more on AX after the talk.
The 2025 Design in Tech Report covers how AI is changing design work, including task-completing agents, the move from UX toward AX, emerging AI UX best practices for trust and usability, and governance risks in automation. It is aimed at people working in design and technology.
For beginners who want to understand UX research and run a first user interview or usability test. You will be able to choose a suitable method, plan a short study and make sense of what you find.
For intermediate designers who want to take a UI from layout grid to working code. You will plan interfaces, build interactive components, document them for developers and use AI to generate screens from a design system.
For researchers who already run studies and want sharper interviews and tests, plus ways to get findings acted on. You will be able to plan, run and present research that shapes product decisions.
For intermediate UX and product designers who want to use Claude and Figma's agent in a consistent way. You will learn to build repeatable design workflows and choose AI tools sensibly for your own process.