I distinguish using AI in my workflow from designing a product that contains AI. The latter also requires decisions about how people understand, review and correct the system’s output.
Choose a useful task
I favor tasks where AI helps compare alternatives or reduces repetitive production work. I review the source material and what can appropriately be shared. Claims about user behavior still need a traceable basis in research or observation.
Direct and refine the output
I work with structured prompts, visual references and revisions. I select and edit the results for relevance, accuracy and consistency with the product’s identity, then inspect them in the layout or interaction where they will be used.
Design review and correction
When AI is part of a product, I start with the user need and the consequences of an unreliable response. I design understandable system states, opportunities to review or correct output and a next step when the system cannot help.
Evaluate uncertain and failed outcomes
I review accessibility, responsive behavior and interaction states alongside the output itself. For AI features, I include ambiguous requests, unsuitable responses and failure paths. I state which behaviors a prototype has and has not tested.
AI contributes to exploration and production. I remain responsible for checking the result and explaining the design choices I keep.
About
Me
I bring together product thinking, visual identity and interaction design, with close attention to context, clarity and the details that make an experience feel considered.
Read more