Designing for AI

If Every AI Feature has a Text Box, How Do You Get It Right? (part 2)

Once chat is ruled in, the craft is composition, editing, and repair loops, not the chat box. How to build AI conversation surfaces that actually hold together.

Every AI Feature Doesn’t Need a Text Box (part 1)

Chat is often the lazy default. Instead, use a pattern selection model that maps task type to the right interaction affordance for AI product design.

Teammate or Tool Call? A Decision Rule for Personifying Agents

Personification is a design decision with real implications. A four-question decision tree that routes every agent surface to its lowest sufficient posture.

The Mental-Model Reset: Designing Systems That Decide

Moving from deterministic screens to probabilistic behavior is the biggest design shift since mobile. A before/after worksheet to reorient your team’s assumptions.

The Agent is the Interface

When software acts on a user’s behalf, the interface stops being a surface and becomes a relationship. That changes what we design.

Trust Is the Material You’re Actually Designing

Every AI/UX decision either deposits or withdraws trust. How to treat trust as the primary design substrate, not as an additive feature, but as the thing every surface runs on.

AI Adoption Is an Ops Problem, Not a Tools Problem

The bottleneck at scale isn’t which AI tool you buy. It’s distribution, standardization, and governance. Design has solved this problem before. We called it a design system.