counterintuitively, the steepest gains from AI in medicine will land in the specialties we describe today as "more art than science." the data-rich fields only look like the frontier because they are easiest to see. the first wave of clinical AI went where objective measurement already existed: radiology, pathology, ophthalmology, etc. the progress is tremendous, and the potential is still wildly under-tapped.

"machine eyes" will surface new signal from data we already collect but cannot fully read, and multimodal models will link findings across studies, time, and modality. these fields will get sharper.

what happens when AI commoditizes diagnosis? for decades, diagnosis has functioned as the central bottleneck in medicine: a gateway to treatment, a billing anchor, and a site of epistemic authority.

the history of software in medicine is fascinating.

judy faulkner wrote the first 30 lines of code for Epic in her college dorm room and bootstrapped the business to a deca-billion dollar market leader.

here's the story of UpToDate, the #1 medical reference tool in the world: UpToDate was founded by Dr. Bud Rose, a Brigham-trained nephrologist who authored one of the most transformative texts in nephrology during his time:

"Clinical Physiology of Acid-Base and Electrolyte Disorders"

over the last few years, I've met with thousands of startups raising capital + have supported many portfolio companies doing the same.

here are some best practices for fundraising efficiently that I've learned from rockstar CEOs: 1/ but first, the purpose of fundraising: secure the bag 💰+ find an awesome investor-partner 👊

CEOs evaluate investors + vice versa.

companies share data in exchange for insights, questions, + clear articulations of how investors can support the next phase of growth.