The conversation around AI in medicine has officially shifted from
“What if?” to “What’s next?”
The AI Doctor Will See You Now
In a historic regulatory milestone, the State of Utah has launched a first-of-its-kind sandbox pilot program. For the first time in U.S. history, state regulators are allowing an AI-driven app named Nolla Derm to operate as an end-to-end autonomous clinical practitioner.
This means a software algorithm is legally authorized to diagnose a patient and write a prescription without mandatory, real-time human intervention.
Authorized by Utah’s Office of Artificial Intelligence Policy, this pilot represents a monumental paradigm shift in medical legality, clinical safety, and the future of healthcare economics.
This means a software algorithm is legally authorized to diagnose a patient and write a prescription without mandatory, real-time human intervention.
Authorized by Utah’s Office of Artificial Intelligence Policy, this pilot represents a monumental paradigm shift in medical legality, clinical safety, and the future of healthcare economics.
How the Autonomous Consultation Works
To keep patient safety at the forefront, the pilot operates within highly strict parameters:
- The Specialty: The software is limited entirely to treating mild-to-moderate acne, a field with highly standardized and structured clinical guidelines.
- The Clinical Flow: A patient fills out a 10-to-15-minute medical questionnaire and uploads a five-angle facial scan. The AI maps out the skin, diagnoses the severity, and issues a treatment plan.
- The Direct Output: The AI chooses from a menu of topical medications and routes the prescription directly to the patient's local pharmacy.
- The Price Point: Disrupting traditional medical costs, the service is offered as a subscription for $4.99 per month.
Graduating to Full Autonomy:
The Phase-Gated Guardrails
Regulators are not giving the AI free rein on day one. The software must prove its clinical accuracy through a strict, data-driven "sandbox" rollout:
- Phase 1 (First 100 Patients): Every single AI-generated prescription must be manually reviewed and cross-approved by two licensed human doctors before leaving the system.
- Phase 2 (Next 400–500 Patients): The AI issues prescriptions directly to pharmacies, but human physicians audit and review the cases on a daily retroactive basis.
- Phase 3 (Full Execution): The system goes fully autonomous, with human oversight dropping to a random sample audit of 10% of cases per week.
To advance through these stages, the AI must maintain a staggering 95% diagnostic agreement rate with human doctors and produce zero serious adverse events. Any complex, severe, or pregnancy-related cases are instantly weeded out by the algorithm and escalated to a 24/7 human telehealth network.
A Mass Expansion of Healthcare Automation
This is not an isolated experiment. Utah’s "pro-human AI" initiative also approved August AI to run a $4 AI-assisted prescription renewal pilot for chronic diseases like diabetes and cardiovascular illness. While August AI’s model still requires a human doctor to sign off on the final renewal, it highlights a broader push toward automated triage.
To ensure absolute transparency, Utah has partnered with elite external auditing bodies, including the Coalition for Health AI (CHAI) and Stanford’s Clinical Excellence Research Center, to independently evaluate the clinical data.
The Professional Debate
Utah’s sandbox experiment sits at the absolute center of a massive global debate.
On one hand, tech innovators and healthcare economists view this as the ultimate solution to medical deserts, weeks-long specialist wait times, and soaring healthcare costs. On the other hand, clinical traditionalists question whether removing the human touch risks missing systemic health issues or over-simplifying patient care.
One thing is certain: the regulatory barrier has broken. The question is no longer whether an AI can legally act as a doctor, but how quickly this model will expand to other fields of medicine.
Created with
