I take on a small number of projects. Most start with one of these questions.
How should research work here?
Centralized or embedded, built in house or bought, and how to show leadership what the function is worth.
How do we study an AI product?
Standard UX methods miss a lot when a product gives different answers to the same question. I help teams design evaluative research and find out how much users trust the system and how well that trust is calibrated to its accuracy.
Can we trust these numbers?
Survey design, sampling, weighting, and knowing when a question needs a regression model and when it needs an hour of listening.
- AI health platform Read the case study → Reshaped the go-to-market model, took adoption from fewer than 20 users to over 16,000, and led to the platform's first six-figure service agreement.
- Tokenized asset platform Read the case study → Research that changed transaction labels and onboarding, and cut compliance approval time from 6 to 8 weeks down to 3 to 5.
I'm a mixed methods researcher and strategist with a PhD in psychology and quantitative methods. I write Ground Truth, a newsletter on research rigor and the gap between how confident a finding sounds and how accurate it is.
I also built The Capability Question, an AI research design tool, and an open source tool that checks whether an AI system's confidence matches how often it's right.