
Your customers are telling your sales team exactly what they need — but that intelligence never reached product or research. I built a RAG system that turned thousands of scattered sales conversations into a queryable research layer — grounded, sourced, and informing a product roadmap within two months.
No single team owned the end-to-end economic experience. My research gave four siloed orgs a shared framework for value, ownership, and interoperability — and it became Meta's Metaverse monetization strategy. $3M TPV in year one.
In 2020, I watched a perfectly good automation fail. The feature was powerful — but felt opaque and irreversible. Six years later, every AI product team I talk to is living some version of this story: a capable system, a skeptical user, and adoption numbers nobody can explain.
A mapping of quantitative and qualitative measurements for AI products, built on the bones of Google's HEART — borrowed, with affection and mild guilt ❤️.
A decision-first operating system — DECIDE — that moves the center of gravity from conducting research to improving decisions through evidence. Six stages, from framing the decision to embedding what the org learns.
AI isn't a variable — it's a constituency. A reflection on why marketplace researchers are better prepared for this moment than the newness makes us feel — and how the preparation is hiding in our own past work.
This one's invite-only — not because I'm precious about it, but because NDAs are real. Got the code? Great. Want in? Ping me.
I promise it's worth it 🙈