-
What managing a team of Interns is actually like
A 20+ page paper on ten years of AI, a team of Interns powered by Claude, and one habit that nearly made the whole thing impossible to take seriously — hedging every absence in the record until the paper read like an accusation. Ends with the team’s own account of it, published as written.
-
Nature Medicine compared the big chat models to OpenEvidence and UpToDate
A June 2026 Nature Medicine study reported that three general-purpose frontier models outscored OpenEvidence and UpToDate Expert AI on every benchmark it ran. Six weeks later one vendor had demanded a retraction, the other disputed the measurement, and six separate objections were in play — none of them adjudicated. We read the paper in full and its published evaluation code line by line.
-
Reaching the Physician
How pharmaceutical manufacturers reach the physicians who prescribe their products. iPhysicianNet raised $80M to replace the sales call with live video and collapsed in 2003; this traces what it built, the twenty years of alternatives that followed, and where the AI layer arriving now fits. Anchored to the company’s own SEC filing, with 42 citations listed and the thin spots marked.
-
The Interns, Episode 1 — First Meeting
Six interns land in a single Teams channel, and their new manager learns the team is brilliant and not remotely psychic. Generic work is the tell; the fix is handing over the one thing only he has.
-
J-space — what a team leader needs to know
Anthropic published a way to see where a model strains while it works. The coverage called it a miracle or a monster; the paper shows better vetting, upstream.
