On October 5, 2026, Nolla Health launched a pilot project in Utah in which artificial intelligence issues initial prescriptions for acne through the Nolla Derm app. The company describes the project as the first case in the US where a state regulator has allowed AI to write an initial prescription specifically: until now, AI in medicine has mostly been limited to refills, while initial prescriptions have always carried a human physician's signature.
The new project is built on an agreement with Utah's Office of Artificial Intelligence Policy and the Division of Professional Licensing, developed in partnership with the Utah Medical Licensing Board. In other words, this is not an unregulated commercial experiment but an official pilot program coordinated with the state regulator — prescribing authority is being tested with the participation of state licensing bodies.
How the Visit Works
Utah residents over 18 can participate in the project. A patient downloads the Nolla Derm app and completes a 10–15 minute visit: first filling out a questionnaire, then scanning their face. Once the process is complete, within a few minutes they receive a personalized treatment plan and — if needed — an initial prescription. The service costs $4.99 per month.
The key constraint is that the AI system does not make freeform decisions: it only selects from topical treatment protocols pre-approved by physicians. Cases the system judges uncertain are automatically routed to a licensed physician — the AI does not write prescriptions for them. This way, the AI operates within strictly defined clinical boundaries.
The prescribing process takes just 10–15 minutes and the answer arrives within minutes — but the speed does not come at the expense of safety constraints: the AI still works only within pre-approved protocols.
Three-Stage Physician Oversight
In the pilot project, physician oversight gradually tapers off. In the first stage — the first 100 patients — every prescription generated by the AI is approved by a physician before it reaches the patient. At this stage human oversight is at its maximum: no prescription goes out unchecked.
In the second stage — up to 500 patients — the AI writes prescriptions itself, while a physician reviews each case afterward, retrospectively. In the third stage, samples of the prescriptions written each week are reviewed. In other words, oversight is never fully lifted — only its form changes: from advance approval to subsequent analysis.
Another important aspect of the oversight model is its openly staged structure: for the first 100 patients every prescription goes through physician approval, for the next 400 patients retrospective analysis is applied, and after that the project switches to a regime of regular sample reviews. This approach is designed to catch AI system errors at an early stage.
The project's scope is tightly limited: topical acne treatments only. Oral medications and isotretinoin are not included in the pilot. This constraint is aimed at minimizing risk — topical agents are considered lower-risk than drugs with systemic effects.
Clinical Evidence and Safety
According to Nolla Health, in real-world practice physicians agreed with the AI's recommendations in over 96% of cases. The remaining cases required only minor adjustments — such as tuning the strength of a topical agent — and no disagreements were observed about the treatment approach itself. In other words, the treatment direction chosen by the AI almost always matched the physician's conclusion.
The 96% agreement rate is presented as the primary measure of the system's clinical reliability. Minor adjustments in the remaining cases — such as tuning topical agent strength — indicate that the treatment strategy itself was chosen correctly.
The AI system operates on structured clinical data and facial scan analysis and has no free-text interface. This is a technical constraint to keep the system from going off-script: the AI makes decisions only within pre-defined protocols.
“There's been a lot of noise about AI in healthcare, and a lot of it deserves scrutiny. What sets Nolla apart is the clinical rigor behind it. The AI isn't making freeform decisions. It's selecting from pre-approved protocols reviewed by physicians, and clinicians are built into every stage.”
— Dr. Zaid Fadul, the project's supervising physician (Nolla Health official blog)
More details on the project are in The Verge's coverage. For now, the program remains a pilot project limited to the state of Utah.
