Most of what I've written about Prometheno so far has lived on Medium — two pieces last summer about why healthcare AI keeps stalling, then almost a year of building, mostly off the page.
(Why the models work but the system still fails · Why AI in healthcare keeps stalling)
This blog is where the writing comes back, and a few things change.
The first is the venue. Medium was a fine starting point: instant audience, clean editor, no infrastructure to maintain. But the writing was always one click away from being trapped behind a paywall or buried in a recommendation algorithm I don't control. Build-in-public means readers, but it also means durability — posts you can link to in three years and trust will still be there, in the form you wrote them. Owning the venue is the easiest way to guarantee that.
Future posts land here first. The earlier Medium pieces stay where they are for now; I'll port them over when I have time.
The second is the cadence. Roughly one piece every two weeks for the next three months, then I'll re-evaluate. Pieces will be smaller than the Medium ones — build notes more than essays. What we shipped, what broke, what I changed my mind about. Less manifesto, more workshop.
The third is what I'll actually write about. The short list:
- What we ended up building after the 1upHealth pivot — what works, what's still painful
- A protocol called HAVEN that came out of the work, and the parts of it I'm least sure about
- Patient data economics: the gap between "fair share of value" as a slogan and "fair share of value" as a number
- Things I got wrong in the Medium pieces a year ago, and what I learned
If you'd like an email when new pieces land, there's a form below.
No tracking, no upsells. RSS at /blog/rss.xml if you'd rather.
— Ziyuan (Chester) Guan