Artificially Designed
A letter on designing AI systems for the humans who use them. It is written for clinicians, analysts, program leads, and founders who have to make a tool work inside a real workflow with real oversight, not in a demo.
The promise is simple. You do not need every tool. You need the right one.
First and third Monday of the month
Two issues a month, both on a Monday. Long enough between them that each one has something tested behind it, close enough together that an open call still reaches you in time. Nothing on this site is gated except the subscribe form itself.
What is in every issue
Built Wrong
One system that looked right on the demo and fell apart in practice, and the specific decision that caused it.
One Slot
A single tool held up against the job it claims. Every job gets one slot, so something has to lose to win it.
Before You Ship
The short list of checks that catch the failures nobody wants to explain to a compliance officer later.
Open Calls
Workshops, meetups, and podcast recordings that are open to anyone reading, with no gate.
One Ask
One question back to you. Answers shape what gets tested and written next.
Prefer to read a few issues first? Browse the archive on beehiiv ↗. The reference material the letter points back to lives in the resources library and the tool calls live in Last Tool Standing.