i make things by talking to machines.

i stopped writing code. i started describing what i want.

somewhere in the last year the tool changed shape under my hands. i used to live in an editor. now i think out loud to a machine and watch the thing get built, corrected, shipped — a conversation instead of a keyboard. it sounds like a gimmick until you've done it for months and realize your job became taste, judgment, and knowing what to ask for.

the people theorizing about agentic development and the people living it are about to be very different. i'd rather be the second kind.

it's also humbling. when the bottleneck is no longer typing, you find out fast whether you actually knew what you wanted. mostly i didn't, and the machine taught me to find out.

i build the plumbing capital runs on.

for years now, mostly the same stubborn problem from different angles — the unglamorous infrastructure underneath raising and moving money. portfolio systems, market mechanisms, and lately tokenizing real-world assets so capital can actually form and settle onchain.

i think the future is markets all the way down — coordination, governance, even truth, priced and cleared by markets we mostly never look at. and the part nobody wants — compliance, custody, the 2am edge case — is exactly the part i find worth doing. increasingly it's the part agents can carry too.

denver rearranged me — altitude and the medicine both.

i came here to build and ended up taking the inner work as seriously as the technical kind. a few years deep in ceremony and psychedelics, the mountains, long stretches of sitting with things much larger than language. it's not a hobby and it's not a personality — it's the part of my life that keeps the rest honest.

i stopped believing ambition and meaning were opposites somewhere on a mountain, probably mid-whiteout.

i think the best systems and the best states of mind share a quality: they get out of their own way. that's as true of a market as it is of a nervous system. i build, and snowboard, and occasionally come back from somewhere far out with a clearer head, toward that.

1. Markets all the way down On prices as compressed belief, coordination as the real game, and why I've come to think most of the world's hard problems are market-design problems wearing a costume. Jun 12, 2026 →
2. Turning the volume back up I quit weed. Not as willpower, not as virtue, because it was turning the volume down on everything, and I wanted to feel my own life at full signal again. A few days in, written while it's still raw. Jun 9, 2026 →
3. Tokenization is mostly compliance (and that's why it's hard) Bringing real-world assets onchain isn't a crypto problem. It's a compliance, custody, and trust problem with code bolted on, the exact grunt work that's resisted automation, and the exact reason it's a moat. Jun 8, 2026 →
4. The day software learns to do the paperwork Agents quietly crossed a line, they can hold a goal and reach for a tool. Paired with back-office finance, that's the thing that's blocked automation for decades finally getting a participant that doesn't sleep. Jun 1, 2026 →

all writing →