AI design must lead with customer need, not technology¶
Insight: The biggest trap in AI product design is defaulting to chat interfaces because ChatGPT, Copilot, Gemini, and Claude are all chat-centric. AI's power isn't in chatting — it's in inference and reasoning. Designers should ask: would chat reduce friction or create new pain points? Does the product even need AI? The most important question remains: am I leading with technology or customer need? Designing for AI is, in that sense, just like designing for any other surface.
Detail: Written by a Microsoft AI product designer. Key design principles: (1) return to research fundamentals — understand how customers work now, (2) design for non-determinism — AI output varies, so build guardrails and confidence signals, (3) integrate AI into the fabric of experiences rather than bolting on chat, (4) think holistically about onboarding, retention, and the full service ecosystem. The framing positions AI design not as a new discipline but as the same discipline applied to a new medium.
Sources
- Will Brennan — "Approaching design in the age of AI" (2025-09-01)