Doctronic: Utah Lets AI Renew Prescriptions Without a Doctor
Ethics & AI 5 min read

Doctronic: Utah Lets AI Renew Prescriptions Without a Doctor

Aisha Patel
Aisha Patel
Apr 8, 2026

On January 6, 2026, Utah became the first state in the U.S. to let an AI system autonomously renew prescription medications — no doctor in the loop at the moment of decision. The pilot program, run by health-tech startup Doctronic, covers 192 drugs for chronic conditions like hypertension, diabetes, and depression.

This isn't an AI that assists a physician. It's an AI that replaces one specific function a physician performs — and it's doing so with the state's blessing, inside a regulatory sandbox that deliberately relaxes existing laws.

Three months in, the experiment is raising questions that go far beyond Utah.

What Doctronic's AI Actually Does

Prescription renewals account for roughly 80% of all medication activity in the U.S. healthcare system. Most are routine: a patient with stable blood pressure needs their lisinopril refilled. Traditionally, that requires a physician to review the case, sign off, and send the renewal to a pharmacy. Delays are common, and the consequences are real.

Medication non-compliance — often caused by prescription lapses — drives over $100 billion in avoidable medical expenses annually, according to Doctronic co-founder Dr. Adam Oskowitz.

Doctronic's system works like this:

  1. A patient initiates a renewal through Doctronic's platform
  2. The AI performs a clinical evaluation — reviewing the patient's history, current medications, and condition stability
  3. If the case is straightforward, the AI autonomously approves the renewal and sends it to the pharmacy
  4. Complex cases get automatically escalated to a human clinician

The AI currently handles renewals for 192 medications — all for chronic, stable conditions. It doesn't write new prescriptions, adjust dosages, or treat acute illness.

The Regulatory Sandbox Approach

Utah's Department of Commerce approved the program under its regulatory mitigation authority — essentially a sandbox that lets companies test innovations under relaxed rules while the state monitors outcomes.

The safeguards are specific:

  • Human physician review for the first 250 patients processed by the AI
  • Automatic escalation protocols for anything the AI flags as complex
  • Contractual prohibition on Doctronic using patient data for other purposes
  • Mandatory disclosure — every patient must be told they're interacting with an AI, not a doctor

The state is tracking medication refill timeliness, patient adherence rates, safety outcomes, workflow efficiency, and cost impacts. Findings will be shared publicly.

What the Critics Are Saying

A Stanford Law School analysis published in March 2026 called the program "a measured step into autonomous medicine" — but identified several serious gaps.

Limited pre-deployment evidence. Stanford's researchers noted that there are no plans for independent evaluation after launch, and Doctronic hasn't committed to publicly sharing its own performance data beyond what the state requires.

Liability questions. Doctronic's terms of service disclaim liability for system errors. If the AI renews a medication that harms a patient, the accountability chain is unclear. Who's responsible — the AI company, the state that approved the sandbox, or no one?

Loss of clinical touchpoints. When a prescription expires, that's often the moment a physician catches something new — an unrelated symptom, a weight change, a blood pressure trend. Automating renewals could eliminate these "incidental screening" opportunities, particularly for patients who don't see their doctor regularly.

The FDA question. Doctronic has publicly stated that its AI system doesn't require FDA approval. Experts disagree. A STAT News investigation in February 2026 found that the program is testing the boundaries of the FDA's authority over clinical AI products — and federal regulators may eventually weigh in.

The Case For Autonomous Renewals

Despite the concerns, the argument for AI-driven renewals is straightforward: people die because prescriptions lapse.

For patients in rural areas, those without reliable transportation, or anyone navigating a fragmented healthcare system, getting a routine refill approved can take days or weeks. The current system introduces friction into what should be the simplest part of medicine — continuing a treatment that's already working.

Doctronic co-CEO Matt Pavelle frames the system as a collaboration tool, not a replacement: patients, pharmacists, and physicians working more efficiently, with AI handling the administrative burden that slows everyone down.

If the pilot's data shows improved adherence rates and no safety regressions, it will be difficult for other states to ignore the results.

The Bigger Picture: Who Regulates AI Medicine?

Utah's experiment exposes a regulatory vacuum. The FDA regulates medical devices and drugs. State medical boards regulate physicians. But an AI system that makes clinical decisions without being a "device" in the traditional sense and without being a licensed practitioner? It falls between jurisdictions.

The regulatory sandbox model gives Utah flexibility, but it also means the safeguards are negotiated, not standardized. What happens when other states adopt similar programs with fewer protections? What happens when the sandbox period ends — does the AI keep prescribing, or does it need formal approval?

These aren't hypothetical questions. Multiple states are reportedly watching Utah's pilot closely, and Doctronic has signaled interest in expanding.

The Bottom Line

Utah's Doctronic pilot is the first real-world test of autonomous AI in clinical decision-making in the United States. The safeguards are reasonable for a pilot. The need is genuine — prescription delays harm patients every day. But the gaps in independent oversight, liability frameworks, and federal regulatory clarity mean this experiment is writing rules in real time. The outcome won't just determine whether AI can refill your blood pressure medication. It will set the precedent for how deeply AI can embed itself into the practice of medicine — and who gets to decide when that's safe enough.