Utah lets an AI app write acne prescriptions with shrinking doctor checks
Startup Nolla Health will scan your face and prescribe acne medication through an app — with physician review that loosens as more patients go through the pilot.
What happened: Healthcare startup Nolla Health has launched a pilot in Utah letting its app scan a user's face, judge how severe their acne is, and write a prescription without a doctor drafting it. Nolla says it's the first in the US to have AI issue brand-new prescriptions, not just renew existing ones. The service costs $4.99 a month and is open only to Utah residents 18 or older with mild-to-moderate acne. The AI can currently prescribe eight different skin treatments.
Why it matters: Utah started allowing AI to renew certain medications earlier this year, and several companies have tested that. Writing an original prescription is a bigger step: it means software, not a clinician, is making the first call on what drug someone takes. Nolla frames this as freeing dermatologists to spend time on harder, riskier cases instead of routine acne — a real access and efficiency argument, especially for people who struggle to get a timely dermatology appointment.
How it works, plainly: A user photographs their face in the app, the AI grades acne severity, and it either selects a treatment or, if it can't confidently decide, routes the case to a physician. Human oversight is heaviest at launch and eases over time: the first 100 patients each get a physician sign-off before their prescription is issued. From roughly 101 to 500 patients, physicians check prescriptions only after they've already gone out. Past that, Nolla says doctors will review at least 10% of prescriptions monthly, plus any case flagged for an escalation or side effect.
The rollout: For now this only applies in Utah, where state rules have opened the door to this kind of AI prescribing, and only to adults with mild-to-moderate acne — not severe cases or minors. Nolla describes the shrinking review schedule as a deliberate pilot design, not a permanent end state, but it also means the proportion of AI decisions actually checked by a human drops sharply as the patient count grows.