Workflows to Think, Test, Build and Ship with AI

I joined Xinran Ma for Season 1 of AI-Powered Design Workflows. My session was a practical workshop on a four-part framework for integrating AI into the design process: Think, Test, Build and Ship. The focus was not on collecting more tools, but on choosing the right kind of assistance at each stage of the work.

Thinking comes first, and manually: work the problem until you have an opinion, then hand Claude the brief and constraints to find assumptions, contradictions and gaps. Testing means throwaway HTML prototypes; one idea went through 58 versions, iteration gets cheaper, judgment does not.

Building moves into Cursor and the production codebase, starting with low-risk fixes that grow technical fluency and engineering trust. Shipping compounds through Watchtower, one queryable system for research, prototypes and decisions, so context outlives the project.

The broader message: AI should increase the quality and reach of design thinking, not replace it. The tools accelerate the movement between thought and implementation; they do not supply the judgment that makes the work valuable.

Key themes: AI-assisted design, product thinking, rapid prototyping, technical fluency, designer–engineer collaboration and knowledge management.

Watch the session here.