Notes on building software
where being wrong has consequences.

I have spent the last fifteen years building things that other people use to do consequential work — first as a founder of a custom-fabrication studio that put twenty-foot interactive sculptures in the lobbies of Hilton, Hard Rock Cafe, and Uber, then as a product leader inside three AI companies of varying scale.

What links those two lives, for me, is the same problem in different costumes: how do you build something that an institution will trust enough to put in front of its customers, its regulators, or its city council? The answer is almost never a feature. It is almost always a posture.

I write because the discipline of writing forces me to figure out what I actually believe. Most of what is here is an attempt to convert pattern-recognition into something I can hand to the next person who has to do this job.