Nolla Health wins first US regulatory nod for AI that writes initial prescriptions

Nolla Health has become the first organization in the United States, and possibly worldwide, to secure regulatory approval allowing an artificial intelligence model to issue initial prescriptions without a physician in the loop. Until now, medical AI has been limited to answering questions, summarizing information or assisting clinicians; a few entities have won clearance to renew prescriptions a doctor originally wrote, but none had received a mandate to initiate drug treatment from scratch.

The clearance follows collaborative work with dermatologists and researchers from Harvard, MIT and Johns Hopkins. The application, Nolla Derm, is already in use, though the company has not published clinical performance metrics or comparative safety data against human physicians. The "end-to-end AI physician" label the founders use describes automation of the routine care pathway — diagnosis, treatment decision, prescription and follow-up — but it remains unclear what control and oversight mechanisms sit outside the model.

The initial launch targets acne treatment in Utah, where roughly 100,000 residents have active acne and 90% have never seen a dermatologist. Utah leads the United States in melanoma incidence, and the state's dermatologists are stretched between routine cases and complex, high-risk ones. Nolla's logic: free physicians from the steady stream of routine work so they can focus on the difficult cases.

The company says it intends to expand into women's health and other specialties, with the stated goal of automating routine care while preserving the quality and follow-up a patient expects from a doctor. What is missing so far: transparency on the model architecture, prospective validation results, and a protocol for handling cases the model flags as anomalous. Until those are published, "end-to-end AI physician" remains more marketing slogan than established clinical definition.