I design for the gap between
model trustworthiness and user trust
For
I've spent the last decade designing products in fintech, healthcare, and consumer tech. I currently lead design for AI products at SoFi, where a wrong answer at the wrong moment can cost someone real money. It's shaped how I think about trust: it isn't earned by making the interface feel confident, it's earned by making the system honest about what it knows. Most of that work lives in the small decisions, type weight, hedging copy, where a verify affordance sits, when the UI should branch instead of commit.
I prototype in code, usually with Claude, and I write at Slow Signal. It's where I slow down on the AI design questions that deserve actual thinking, not takes.
How I build
Three design principles that shape how I approach AI products, from first prototype to production.
Making trust track capability is not an abstract principle. It shows up in timing, typography, and copy, and it requires the same detail attention as any other craft surface.
Stale point estimate → Range with data provenance