Articles

Designing interfaces for agents people can understand, steer, and trust.

Writing on agent-native product design, interface behavior, and the decisions that turn capable models into products people can confidently use.

July 23, 2026

A high-fidelity prototype needs a decision trail

A production-like prototype shows what an experience does. A decision trail explains why it works that way, what still needs testing, and where reviewers should focus.

Read article

June 28, 2026
Users don't blame the model. They blame you.
Most AI teams obsess over output quality while users abandon during the wait. The 30 seconds between click and result is the highest-stakes UX surface in your product — and nobody designed it.
Read article
June 28, 2026
The agency that built your AI product moved on. That's the real design risk.
When the agency that built your AI product moves to their next client, the design intent walks out the door with them. Models silently update, behavior drifts, and nobody on your team knows what 'working correctly' was supposed to mean.
Read article
June 16, 2026
Your AI agent can read your codebase. It can't read your mind
Code captures roughly 40% of a product's identity — the other 60% lives in design decisions, interaction principles, and things never written down, which is exactly why agentic coding tools produce generic output.
Read article
June 14, 2026
Standing up a Storybook from scratch with AI
A concrete how-to for the Shadcn + Storybook + Cursor flow: how to go from designs to matched, documented components without manually writing the boilerplate.
Read article
June 10, 2026
The spinner is a lie
Loading patterns built for deterministic systems fail completely in agentic contexts — silence, loops, and dead ends have no visual language yet, and that's a design problem nobody's solving.
Read article
June 10, 2026
Lo-fi wireframes from Claude: the design thinking shortcut nobody talks about
Using Claude to generate rough layout options mid-design isn't about output — it's about thinking faster without having to build your way to a bad idea first.
Read article
June 8, 2026
The activity panel isn't optional. It's the product.
One stream can't do two jobs. Most agent interfaces force it to anyway — and that single non-decision is why your product feels confusing even when the model is good.
Read article
June 8, 2026
Nobody calls an agency when the building is on fire
The project-based agency model hands you a polished snapshot and walks away. AI products never stop moving — which is why they need continuity, not deliverables.
Read article

A high-fidelity prototype needs a decision trail

More articles

Users don't blame the model. They blame you.

The agency that built your AI product moved on. That's the real design risk.

Your AI agent can read your codebase. It can't read your mind

Standing up a Storybook from scratch with AI

The spinner is a lie

Lo-fi wireframes from Claude: the design thinking shortcut nobody talks about

The activity panel isn't optional. It's the product.

Nobody calls an agency when the building is on fire