My AI UX Approach
My process, patterns, and principles for designing trustworthy AI experiences.
My (evolving) AI-augmented process
AI compresses the pass. The judgment, the taste, and the read on what the user actually needs — that part's still mine.
Synthesizing research notes into themes.
Drafting blockflows I refine rather than starting from a blank canvas.
Generating variants I react to and refine — often surfacing options I wouldn't have considered.
The handoff gap between design and engineering shrinks, and I stay closer to how the product actually behaves in users' hands.
The patterns underneath
A few interaction patterns show up again and again once you're paying attention — not tied to one product, but to the stage of the exchange between a person and a system. Adapted loosely from Aiverse's five-phase framework for AI design, here's the shape I watch for, and where I've actually seen it done well.
Designing for Trustworthy AI
The friction people feel with AI isn't primarily a capability problem. It's a shared-understanding and alignment problem.
Any effective working relationship depends on knowing what's being handed over, why decisions are made, and how to correct course when something goes wrong. Most AI products haven't been designed around those mechanisms yet.
Good delegation isn't built in a single handoff. You establish how you work, what's important, where the boundaries are, and what success looks like. Trust builds through demonstrated understanding.
Most AI products skip this step. They execute immediately and hope the output matches what you intended.
People ask AI to triage email, book travel, or draft reports — then hover over every action and double-check every output.
That's not user resistance. It's appropriate caution for a relationship that hasn't earned trust.
AI products are still optimized for speed to output rather than speed to understanding. Demo culture rewards instant answers, while small cues like "thought for 5 seconds" reinforce the idea that faster means smarter.
But speed doesn't create the shared understanding required for delegation.
So what does trustworthy AI actually look like?
Building the clarity, control, and confidence needed for people to delegate to AI.
Before anyone hands work to an AI system, they need what they'd need from a new hire: an explicit scope of what it will and won't do, which actions require approval, where the checkpoints sit before anything consequential happens, and the zones it won't enter on its own. Most AI products skip straight to execution instead, and ask the user to define boundaries after the fact, one correction at a time.
Once the system acts, the person needs visibility into what it's doing and why — not just the result. That means real-time visibility into current state, the reasoning behind a given choice, clear progress signals, and an honest confidence marker when the system itself is uncertain. An output with no visible reasoning is a black box, however good the output turns out to be.
Refinement only feels safe when it's cheap to get wrong. That means quick undo grouped by task rather than by individual click, clear restore points, high-stakes actions staged in phases instead of executed in one shot, and honesty about what recovery actually costs. Without that, every refinement carries the weight of a decision that can't be taken back — so people stop refining and start avoiding.