Who It's For
HUMANS FIRST
Augmenting humans, not auditioning to replace them
The most overhyped AI promises to make people unnecessary. The most useful AI makes them formidable. I start every design from the human still in the loop — which calls stay theirs, which judgment the system should defer to, and where automation has genuinely earned its keep versus where it's just showing off. Turns out the systems that respect human judgment also tend to be the ones that actually work. Funny how that goes.
LEADERSHIP
Getting tribes that don't trust each other into the same room — and out aligned
The best AI work happens at the intersection of people who'd rather not be intersecting: engineering, product, risk, legal, and a C-suite that wants the upside without the asterisks. I'm fluent enough in each dialect to translate, blunt enough to surface what everyone's avoiding, and diplomatic enough that they're still speaking to me afterward. The goal isn't consensus theater — it's turning a fragmented set of stakeholders into a team that actually ships the thing.
THE LONG VIEW
The public-interest mindset: the long view, the right view
Some of the most consequential work in technology happens where interests refuse to agree — privacy against utility, speed against safety, what's profitable against what's right. I don't pretend those tensions resolve themselves; I sit in them. I take the long view even when the short one is louder, and the right view even when it's the inconvenient one in the room. AI is going to shape decisions far beyond the people building it, and someone in the room should be accountable to the people who never asked to be in the dataset. I try to be that someone.