AI Product Design
AI product design is the practice of shaping how people understand, trust, and control AI-powered features—not only how they look.
It spans capability disclosure, input and output affordances, failure states, human oversight, and feedback loops across chat, agents, and multimodal surfaces.
What it means
Designing the full experience around probabilistic software: what the model may do, how users verify results, and how the product recovers when the AI is wrong or uncertain.
Why designers should care
Most AI product failures are UX problems (unclear limits, weak trust cues, bad handoffs), not model quality alone. Shared vocabulary helps teams spec and review AI features consistently.
Example
A copilot drafts code in a diff view with accept/reject per hunk, cites the files it read, and shows a clear message when it cannot access a repo path.
Common mistakes
- • Shipping a chat box without defining success, failure, or escalation paths.
- • Treating AI design as visual polish on top of a generic assistant.
- • No difference between low-stakes suggestions and irreversible agent actions.