If you want to understand how the American professional class imagines the future of the working person, you could do worse than look at a nondescript office building on East 300 South in Salt Lake City. There, inside the Utah Department of Commerce, sits an entity called the Office of Artificial Intelligence Policy. Its bureaucratic charter sounds like something concocted in a business school seminar room: run a state-sanctioned regulatory learning laboratory where private software firms can test unvetted technologies on ordinary citizens without the nuisance of established public protections.
On September 22, 2026, the office finalized a document known as a Regulatory Mitigation Agreement. Under its terms, the state officially waived enforcement of state medical licensing statutes, codified in Title 58 and Utah Code Section 58-17b, for a New York–based tech startup called Nolla Health. In doing so, Utah regulators granted an algorithm the legal authority to diagnose patients and issue first-time drug prescriptions directly to commercial pharmacies without any contemporaneous review by a human doctor. The program launched publicly on October 5, 2026.
The official packaging for this deregulation is nothing short of breathtaking. Regulators and tech founders did not present the maneuver as a corporate giveaway, nor as a desperate bid by venture capital to monetize health care. Instead, they draped it in the language of egalitarian populism. Handing prescription pads over to automated code was christened an act of democratizing care—a philanthropic crusade under Utah’s official “Pro-Human AI Initiative” to rescue rural and lower-income families from six-month specialist waits and unaffordable medical bills. If you cannot afford an afternoon off work and a three-figure specialist copay, the technocrats announce, you should be grateful to point your smartphone camera at your face, pay a monthly software subscription, and let an algorithm mail you pills.
This is the classic formula of modern corporate liberalism, executed with western libertarian efficiency. First, allow an unregulated market to systematically strip basic social provisions from working-class communities, closing rural clinics, starving county health systems, and turning specialist care into an extortionate luxury. Then, step forward with an app, proclaim the resulting degradation of standards to be an egalitarian triumph, and instruct the working class to applaud its own abandonment.
The Silicon Valley Assembly Line
The mechanics of the Nolla Health experiment read like an efficiency consultant’s dream of Taylorist factory production applied to the human body. To obtain a prescription, a patient opens the Nolla Derm application on a mobile phone, completes a standardized digital questionnaire, and uploads a five-angle scan of their face. From those digital pixels and survey checkboxes, the company’s proprietary software determines a clinical diagnosis for acne and selects from a menu of eight topical prescription formulations.
Under the mitigation agreement signed with the state, this process rolls out in three distinct phases designed to systematically eliminate human labor from clinical judgment:
- Stage 1: For the first one hundred patients, two Utah-licensed physicians must independently review and sign off on each automated prescription before it is transmitted.
- Stage 2: For the next batch of up to five hundred patients, the human physician is removed from the loop prior to dispensing. The algorithm transmits prescriptions directly to retail pharmacies, while doctors are relegated to a retrospective review of patient files once a week.
- Stage 3: In the full production stage, the software acts as an entirely autonomous prescriber. Human medical oversight is reduced to an audit of an arbitrary 10 percent sample of patient files once a month, along with occasional reviews of self-reported adverse clinical events.
For this privilege, patients are charged an ongoing monthly subscription fee of $4.99 simply to keep the application active. That fee does not cover the cost of the actual medicine. Patients must then buy the company’s proprietary compounded morning-and-evening topical kits for $49.99 a set, or have generic formulations routed to their local pharmacy.
Nolla Health, operated through its corporate parents Magic Health, Inc. and Magic Health Services, LLC, is backed by $4.5 million in seed funding led by the venture capital giant General Catalyst. When announcing the state contract, Nolla’s co-founder and chief executive officer, Luis Wenus, did not hesitate to claim historic ground. He declared his company “the first company in the US (and possibly the world) authorized to let AI prescribe medication autonomously.” Wenus framed the company’s profit model as a social mission, insisting that “AI will matter most in healthcare when it helps people get care they otherwise would not receive.” He pointed directly to working-class financial distress to justify bypassing human doctors, arguing that the service is tailored for those who “cannot access or afford dermatology appointments, which can cost hundreds of dollars, to treat their skin conditions.”
The company’s chief medical adviser, Dr. Zaid Fadul, was brought forward to reassure the skeptical that the assembly line was thoroughly hygienic. “We don’t want to remove doctors from medicine,” Fadul said during the company’s promotional rollout. “The goal is to free up our time so we can focus on patients who actually need our attention.” Fadul defended the autonomous pipeline by asserting that “the AI isn’t making freeform decisions. It’s selecting from pre-approved protocols reviewed by physicians, and clinicians are built into every stage. That’s a standard AI in medicine should be held to.”
Listen closely to that rhetoric, and you will recognize the familiar cadence of elite condescension. The argument rests on the notion that the health care of ordinary working people consists mostly of minor clerical inconveniences—friction to be optimized away by a software pipeline so that actual human doctors can conserve their precious attention for higher-value consumers. Skin conditions, after all, are rarely just skin conditions; they can be the cutaneous manifestations of underlying endocrine disorders, autoimmune flares, chronic occupational exposures, or malignant melanomas that an off-the-shelf camera scan easily misclassifies. Yet to the technocratic mindset, diagnosing a human being is simply an image-classification problem, easily resolved by replacing an expensive, educated labor force with server cycles.
The Physician Revolt Regulators Ignored
The state of Utah did not stumble into this arrangement by accident. It is the deliberate fruit of Senate Bill 149, enacted by the Utah State Legislature in 2024 as the Utah Artificial Intelligence Policy Act. The statute established the Office of AI Policy and created the Artificial Intelligence Learning Laboratory Program. The core purpose of that program is simple: grant executive branch political appointees the power to sign regulatory mitigation agreements that freeze the enforcement of consumer protection laws, licensing standards, and malpractice statutes so that private tech startups can use the public as an uncompensated testing ground.
The Nolla Derm experiment is not the state’s first foray into algorithmic medicine. In January 2026, the office authorized an autonomous prescription renewal pilot with Doctronic, a health tech firm that went on to raise $40 million in venture funding that March. The state allowed Doctronic’s automated software to issue thirty-, sixty-, and ninety-day prescription renewals for chronic metabolic and cardiovascular conditions, including hypertension and type 2 diabetes.
When the state’s practicing doctors discovered that political appointees were handing autonomous prescribing power to Silicon Valley code, the medical establishment responded with open alarm. On April 20, 2026, eleven of the fourteen physician members of the Utah Medical Licensing Board signed a blistering open letter to the Department of Commerce demanding the immediate suspension of the automated prescribing program.
The licensing board warned that the executive branch was playing Russian roulette with patient lives, noting that proceeding with automated prescribing without consulting the state’s clinical oversight body “potentially places Utah citizens at risk and remains a major concern of the board.” The physicians pointed out that statutory prescription renewal is not a thoughtless clerical chore, but a nondelegable clinical responsibility: practitioners must “reassess clinical stability, monitor potential side effects, screen for newly emergent contraindications, and verify drug interactions, lest patients remain on outdated or suboptimal regimens for months.” The board wrote plainly: “We must not allow AI or other financial motivations to override this obligation, yet that is precisely what occurred here.”
The reaction of Utah’s executive regulators to this plea from the state’s own medical board was swift and contemptuous. On April 21, 2026, the Department of Commerce summarily dismissed the doctors’ warnings, refusing to pause the automated pilot. Executive branch officials insisted that statutory regulatory sandboxes legally supersede the jurisdiction of professional licensing boards. The technocrats had spoken: the professional judgment of practicing physicians, accumulated over centuries of clinical science and codified in statutory labor protections, was treated as a reactionary trade union obstruction standing in the path of venture-backed “innovation.”
When the clinical backlash finally forced the state to make concessions, the retreat only proved how reckless the program was in the first place. On October 16, 2026, Zach Boyd, the director of the Office of AI Policy, confirmed to reporting outlet Nextgov/FCW that regulators had quietly been forced to pull two critical pharmaceuticals off the automated renewal roster: flecainide, a potent and dangerous antiarrhythmic heart medication with a narrow therapeutic window, and butalbital, an addictive barbiturate frequently abused in combination headache therapies. Regulators had originally turned over the dispensing of barbiturates and delicate cardiac rhythm drugs to an algorithmic routine, retracting them only after the state’s doctors forced the issue into the open.
“Half of It Is Probably Wrong”
Yet if Utah regulators were chastened by the near-disaster, they showed no sign of it. On October 15, 2026, Boyd appeared at the ADAPT artificial intelligence conference in Santa Fe, New Mexico, to pitch Utah’s deregulatory model to a national audience. His remarks, reported by Nextgov/FCW the following day, offered an astonishing window into the casual improvisation governing this corporate giveaway.
“We hit a nerve,” Boyd acknowledged candidly, reflecting on the fury radiating from Utah’s medical community. He then conceded that the state had launched autonomous prescribing without having the faintest legal idea of who bears responsibility when the software inevitably harms someone. Utah regulators, Boyd revealed, are currently scrambling to draft an emergency “liability bill” to figure out who gets sued when an algorithmic prescription causes systemic toxicity or misdiagnoses a lethal illness.
Boyd’s defense of the state’s official administrative guidance for AI medical oversight was even more revealing: “Half of it is probably wrong,” he told the audience, “but it’s definitely, it’s got to be, about the most mature thing that exists.” He added with cheerful managerial detachment: “One of the underrated advantages of a sandbox is it helps the regulator learn so much.”
One must pause to appreciate the class politics embedded in that single sentence. When state officials decide to “learn” by suspending medical licensing statutes, it is never the partners of venture capital firms or senior state directors whose families serve as the experimental subject pool. The executive class does not send their children to scan their faces on an iPhone to manage chronic conditions through an automated subscription portal; they take their children to private pediatricians and elite specialist practices in university hospital corridors. The people who “help the regulator learn” are the working-class families in the Salt Lake suburbs, the service workers in Ogden, and the agricultural laborers in Sanpete County who cannot afford to take a day off work or scrape together three hundred dollars for an office visit.
Even worse, Boyd openly boasted that Utah’s localized deregulatory bypass could serve as a model to dismantle health regulations nationwide. “It’s really appealing to people to have kind of a vet once, deployed everywhere, aggregate oversight across jurisdictions,” Boyd said in Santa Fe. “A model like that feeds nicely into the federal government and the kind of oversight that they’ll also want to exercise.”
What Boyd was describing is the corporate dream of regulatory forum shopping. Under federal law, the Food and Drug Administration treats clinical decision-support software that independently makes diagnostic or treatment choices as Software as a Medical Device, or SaMD. Federal oversight requires extensive premarket clearance, such as a 510(k) submission, and demanding clinical trials before automated diagnostic tools are loosed upon the public. By creating a state-level “sandbox,” Utah has manufactured an end-run around the Food, Drug, and Cosmetic Act. A startup needs only to find one compliant state commerce commissioner willing to declare their jurisdiction a testing laboratory, and the company can deploy autonomous medicine into the wild, bypassing federal scientific scrutiny entirely.
Margaret Woolley Busse, the commissioner of the Utah Department of Commerce and co-chair of the state’s Pro-Human AI Task Force, issued an official statement on October 5, 2026, insisting that this end-run represents the cutting edge of enlightened public administration. “The Office of AI Policy is creating a model framework where AI can enhance healthcare outcomes while ensuring robust oversight and accountability,” Busse proclaimed. “With these advancements, we are paving the way for a future where AI serves humanity, prioritizing patient safety and public trust.”
This is managerial doublespeak elevated to art. Public trust is not built by ignoring the state’s own licensing boards, nor is patient safety prioritized by handing prescription pads to software whose oversight director openly admits that half of his regulatory framework is “probably wrong.”
The Fintech Precedent
To anyone who has watched corporate power operate across the American West over the last four decades, the script unfolding in Salt Lake City is deeply familiar. It is the exact playbook perfected during the financial deregulation crusades of the 2010s.
In 2018, Utah became the national pioneer of the “fintech regulatory sandbox,” working in close concert with the federal Consumer Financial Protection Bureau under the Trump administration. The rationale offered to the public then was identical to the rhetoric being peddled for Nolla Health today. We were told that strict state usury laws, mandatory credit reporting disclosures, and traditional banking charters were antiquated relics of an industrial past. These rules, tech lobbyists argued, were holding back innovative app developers who only wanted to expand access to banking for the “unbanked and underbanked.”
The high-minded rhetoric about financial inclusion lasted just long enough for the ink to dry on the statutory exemptions. Within years, the fintech sandboxes had spawned an explosion of predatory payday-lending applications, algorithmically hidden junk fees, and buy-now-pay-later schemes that targeted low-income households, trapping working families in cycles of compounding digital debt while insulating corporate executives behind liability waivers. The promises of universal financial empowerment dissolved into simple class predation dressed up in the language of Silicon Valley disruption.
There is, of course, an essential limit to the financial parallel, and it is a terrifying one. When an app-based fintech firm fleeces a warehouse worker through hidden overdraft charges or balloon interest payments, the worker loses money. When an automated prescribing engine misinterprets an inflammatory skin condition, misses an emergent autoimmune crisis, or dispenses contraindicated compounds, the patient suffers physical devastation, organ damage, or death.
We have also seen this dynamic inside medicine itself. In the 1990s, the high priests of the corporate health care counterrevolution unveiled health maintenance organizations and “managed care.” Back then, the technocrats also promised that they were going to rationalize a broken medical system, eliminate administrative waste, and lower costs for working-class families. They dispatched armies of management consultants into hospitals with thick binders of rigid clinical protocols. If an autoworker or a school cafeteria worker needed an MRI or a specialized surgical procedure, an administrative gatekeeper armed with an actuarial flowchart was there to deny the claim and protect the insurer’s profit margins.
That corporate offensive failed to lower health care costs for ordinary families; it merely fueled insurance consolidation, enriched corporate hospital chains, and degraded the working conditions of nurses and physicians. Yet even the most ruthless managed-care regimes of the 1990s retained one indispensable structural safeguard: an actual human physician had to sign their name to the prescription or the denial. Utah’s automated sandbox crosses that historic Rubicon. Under Stage 2 and Stage 3 of the Nolla Derm agreement, the human physician’s signature is eliminated entirely. The corporate algorithm itself becomes the doctor.
The Two-Tiered Settlement
The deepest deception of the Utah experiment lies in the argument advanced by Nolla Health’s executive leadership: that automated medicine is an act of democratic charity for those who cannot afford specialist care. When Luis Wenus argues that his app gives poor patients an alternative to waiting months for a three-hundred-dollar specialist visit, he is taking the criminal failures of the American market-based health care system and weaponizing them to justify lowering clinical standards for the poor.
Why is specialist care in the United States so scarce and so expensive? It is not because the mysteries of dermatology are inherently beyond human reach. It is because private equity firms have spent the last fifteen years buying up independent dermatology and primary care clinics across the country, gutting clinical staff, imposing predatory billing practices, and driving doctors into early retirement. It is because monopoly pharmaceutical cartels charge extortionate prices for basic off-patent therapies. It is because corporate hospital chains systematically close unprofitable rural clinics while opening luxury surgical pavilions in affluent suburbs.
The democratic, populist solution to this crisis is obvious, time-tested, and rooted in the New Deal tradition: enforce federal antitrust laws against health care monopolies, expand Medicare to cover all Americans, publicly fund the training of hundreds of thousands of new doctors and nurses, forgive the student debt of clinicians who agree to practice in underserved and rural areas, and build a universal public health infrastructure accountable to citizens rather than shareholders.
Instead of doing that, the political and corporate establishment has constructed a two-tier medical caste system. For the affluent professional class in the leafy suburbs of Salt Lake City, Boulder, or Silicon Valley, there will still be real human doctors. They will continue to sit across from experienced clinicians who look them in the eye, examine their skin under magnification, take comprehensive medical histories, and carefully weigh the nuances of their physical and mental health. But for the working class—for the truck drivers, the Amazon warehouse pickers, the home health aides, and the rural mill hands—the technocrats offer an automated surrogate. They are told that high-quality medicine is an unaffordable extravagance, but that if they download a proprietary app and scan their faces into a server farm, venture capital will graciously grant them an automated prescription for fifty dollars a month.
This arrangement degrades both the patient and the worker. It treats the complex, human craft of clinical medicine as nothing more than a mechanical sorting task that can be automated away, devaluing the clinical skill of practitioners while isolating patients in an unaccountable corporate web. As Eduardo Dias, a health technology researcher, noted to The BMJ on October 19, 2026: “Deploying autonomous AI to prescribe drugs without clinician oversight is the wrong trajectory for healthcare. Technology should assist professionals rather than replace them . . . Keeping healthcare workers at the centre ensures AI prevents drugs errors and alleviates burnout without compromising accountability.”
When that accountability is removed, the legal system inevitably protects the powerful at the expense of the weak. Under the terms of the sandbox, a patient who suffers an adverse drug interaction or whose misdiagnosed cancer spreads because a smartphone app failed to identify it is cast into a legal no-man’s-land. The startup will point to its terms of service, asserting that the software is merely a wellness tool or an administrative assistant. The reviewing physician, who audits only a random 10 percent sample of files weeks after the fact, will argue that they never had a formal doctor-patient relationship with the individual. And the state regulators, who waived their own licensing and consumer-protection statutes, will shrug their shoulders and remind the public that their sandbox was designed to help them “learn.”
A Pushback from the Front Lines
The good news is that the working clinicians who actually care for patients are beginning to fight back. Across the country, the medical rank and file is waking up to the reality that tech conglomerates view their professional judgment not as a public good, but as an unnecessary labor cost to be engineered out of existence.
On October 16, 2026, the Medical Group Management Association, representing over 350,000 physicians and medical group practice leaders nationwide, took a decisive stand. The organization formally endorsed the bipartisan Doctors Not AI Act, introduced in Congress as H.R. 10210 by Representatives Greg Landsman, Buddy Carter, Kim Schrier, and Tom Barrett. The proposed federal statute directly confronts the algorithmic takeover of clinical care, prohibiting corporate insurers and autonomous software platforms from using artificial intelligence to override human clinical judgment, deny necessary therapies, or autonomously dispense prescription drugs without direct, contemporaneous physician authorization.
The legislative battle lines are now drawn with unusual clarity. On one side stands a bipartisan coalition of venture capital firms, corporate hospital executives, and state technocrats who view medicine through the lens of a quarterly earnings report—people who believe that treating the sick is just another administrative friction to be optimized away by an automated feedback loop. On the other side stand practicing physicians, patient advocates, and working people who insist on an older, more durable truth: that human health is a public good, that clinical judgment cannot be severed from ethical responsibility, and that ordinary citizens deserve the care of real human beings rather than the cost-cutting algorithms of venture capital.
As tech lobbyists fan out across state capitals from Phoenix to Cheyenne seeking to export the Utah model, the rest of the country should look closely at what is happening in Salt Lake City. The choice before us is not whether technology will play a role in modern medicine. The choice is whether technology will be directed by the public to assist human caregivers in a universal, equitable health system, or whether it will be wielded by corporate monopolists to institutionalize a degraded, automated, two-tiered standard of care for the working class.