Nolla Health's AI now writes first-time acne prescriptions in Utah pilot

Nolla Health, a healthcare startup, has launched a service in Utah that lets an AI system analyze a user's face scan and autonomously write a prescription for acne treatment. According to The Verge, which cites an earlier Bloomberg report, the company announced the program on Monday and describes it as the first service in the country to issue initial prescriptions — not just renewals — generated by AI.

How the service works

The product runs through Nolla Health's app: users scan their faces, and the AI evaluates acne severity before deciding on a treatment. Access is deliberately narrow. It costs $4.99 per month, is open only to Utah residents aged 18 or older, and applies to people with mild-to-moderate acne.

Bloomberg reports that the system can currently prescribe eight different skin treatments. If the model cannot confidently settle on an appropriate option, the company says it will route the user to a physician instead of issuing a prescription.

Physician oversight steps down over time

The launch is structured as a pilot with graduated physician supervision. According to The Verge, two physicians will approve every AI-generated prescription before it goes out for the first 100 patients. For the next phase, covering up to 500 patients, doctors will review prescriptions only after they have been issued. Beyond that point, physicians will examine a sample of at least 10 percent of prescriptions each month, along with every case involving an escalation or a side effect.

Nolla Health frames the tool as support rather than replacement. In its press release, the company states that the pilot is designed to complement physicians, and that its dermatology product, Nolla Derm, takes on routine, low-acuity cases so dermatologists can concentrate on complex, high-risk patients.

Utah as a testing ground

Utah has become a focal point for this kind of experiment. The Verge notes that the state began allowing AI systems to renew certain medications earlier this year, and several other healthcare companies are already trying AI-assisted prescribing there. Nolla Health's claimed distinction is that it issues first-time prescriptions rather than renewals — a new clinical decision made by software rather than a repeat of a decision a doctor already made.

Why it matters

That distinction is the heart of the story. A renewal repeats a judgment a human clinician has already made; an initial prescription is a fresh clinical call issued without a doctor in the loop at the moment of decision. If the model performs well on a well-bounded, lower-risk condition like acne, it strengthens the argument for autonomous AI in other narrow medical domains. A visible failure — a missed underlying condition, an adverse reaction, a prescription outside the intended severity range — would give regulators and skeptics a concrete counterexample.

The tapering oversight model deserves attention too. Moving from prospective physician approval to after-the-fact sampling shifts the safety burden from review before a prescription is issued to monitoring after the fact. Once the pilot matures, a 10 percent monthly sample means most prescriptions will never be seen by a doctor. How escalations and side effects are detected and reported will largely determine whether that safety net is real.

For now, the guardrails are tight: one condition, one state, a fixed menu of treatments, and a rule that uncertain cases go to a human. That narrowness is the point. It is the safest possible configuration for a first-of-its-kind deployment, and it offers a template for how autonomous AI in medicine is likely to expand — one bounded task at a time.