Course Description
The interfaces people will use next do not wait to be clicked. AI agents now reason, act, and adapt on a user’s behalf, and the discipline of designing for them barely exists yet. This studio course puts students on that frontier. Working in small teams that mirror real product design groups, students take a concept from pitch to working prototype through user research focused on intent and trust.
Teams will create a customer journey map showing human-agent handoffs, failure states, and recovery, along with an agent persona defining goals, reasoning style, memory, autonomy boundaries, and escalation rules. Students will then build an individual LLM-based prompter and a team Figma prototype, test and iterate with real users, and present a final design proposal grounded in user evidence, UX theory, and the constraints of AI capability. Students leave with a portfolio piece at industry depth and a defensible point of view on what makes an agent trustworthy.
In this course you will:
- Design agentic UX systems and translate human goals into agent behaviors, guardrails, and interaction contracts
- Create agent personas and human-agent journeys that account for trust, handoffs, failure, and adaptation
- Prototype, test, and refine an agentic experience that demonstrates real task execution
- Evaluate AI systems for usability, reliability, explainability, and ethical risk, and defend design decisions to stakeholders
AI use expectations: Required. AI tools are the material of this course. Students work with Figma and Figma Make alongside tools such as Google Gemini, ChatGPT or Claude, and Google Stitch. AI may support brainstorming, outlining, light copyediting, and rephrasing students’ own ideas, but it may not replace their reasoning or produce unchecked work. Students must disclose AI use and remain responsible for the accuracy and originality of all submitted work.
Special considerations: This advanced studio requires COMMLD 511, 512, or 517, or demonstrated familiarity with design thinking and UX fundamentals. In-person attendance and active participation are required. Students work in small teams throughout the quarter on a concept selected by the class. Some tools have daily token limits, and students may need to purchase additional access.

University of Washington