Triadic care is not being deployed into medicine. It is being deployed into the definition of medicine — and the two are not the same intervention, on the same timeline, with the same remedies.
Here is the scenario that closes the governance window: a physician working a rural urgent care shift sees a patient with an ambiguous presentation — atypical chest pain, unremarkable vitals, no obvious red flags. She assesses, treats, documents. The patient deteriorates. Plaintiff’s counsel retains an expert who testifies that, as of 2029, a reasonable clinician in a networked facility would have engaged the AI co-clinician system — that the system flags the precise pattern the physician missed, that this is precisely the documented use case the system was validated for, that multiple specialty society guidelines now reference the system as a component of consultative best practice. The case survives summary judgment on expert testimony and emerging practice patterns. The jury finds for the plaintiff. Within six months, three major malpractice carriers update their underwriting guidelines: documented AI consultation becomes a factor in premium calculation for primary care physicians.
That case does not need to be won on the merits of the technology. It needs only to be plausible — plausible enough to survive summary judgment, plausible enough for defense attorneys to advise every subsequent client to document system use as a matter of routine. At that moment, the optional becomes obligatory, not by regulation but by risk calculus. The liability regime has rewritten the standard of care, and no one voted on it.
This is the mechanism that most analysis misses. The debate around triadic care has focused on data extraction, workforce displacement, and simulation validity — all real concerns, all downstream of the more fundamental structural event. The event is not adoption. The event is the redefinition of clinical obligation. And that redefinition does not require universal deployment. It requires one sufficiently consequential precedent.
The capture is simultaneous, not sequential
The conventional account of technology lock-in runs in a familiar sequence: a system is deployed, achieves market penetration, then lobbies for regulatory entrenchment. Capture follows adoption. This is how most critics frame the triadic care threat — as a future problem contingent on scale.
The sequence does not hold here, and the Epic analogy is instructive. Epic did not capture clinical standards after hospitals adopted it. It captured standards by being adopted — through user groups that became guideline committees, research partnerships that became published outcomes, workflow defaults that became audited requirements. Epic captured documentation and workflow standards, which became audit and reimbursement reality. Triadic care captures decision-support expectations, which become liability reality. The installed base wrote the standards. The standards mandated the installed base. These were not sequential events. They were the same event, observed from different vantage points in the causal chain.
Triadic care is replicating this architecture at speed. The Harvard, Stanford, and Beth Israel partnerships are not neutral validation studies in the scientific sense — though they generate real signals, their institutional role exceeds their epistemic content. They are the first clauses in an evidentiary argument that will eventually appear in a courtroom: leading academic medical centers have integrated this system into standard consultative practice. The NOHARM framework adaptation is not merely a measurement instrument. It is a paper trail — one legible to courts, insurers, and medical boards. The phased global deployment across jurisdictions with varying regulatory thresholds is not scientific rigor. It is precedent accumulation at scale, seeking the fastest path to the volume of deployment that allows the system to be described, accurately, as widely used.
The governance window is not “now” in the sense of pre-deployment. It is the next 18 to 24 months of early deployment — before enough case law and insurer behavior accumulate to freeze the architecture in place.
By the time organizations feel the lock-in pressure, the standard will already have moved. They will not be choosing whether to adopt the system. They will be managing the liability exposure of not having adopted it sooner.
The physician supervision clause is a toll, not a brake
The most persistent misreading of the triadic care architecture is the physician-in-the-loop requirement. It is described, consistently and by design, as the safety mechanism — the human backstop, the red flag detector, the locus of clinical authority. This description is accurate about the formal structure and misleading about the functional reality.
Operationally, the supervision requirement is a compliance wrapper. The AI system generates the clinical substrate — the initial assessment, the differential, the consultation record. The physician reviews and signs. That signature is not primarily a clinical act. It is a billing act, a liability act, a regulatory act. It converts AI-generated output into a reimbursable medical service and absorbs the malpractice exposure that would otherwise fall on the platform provider.
The arbitrage is not subtle once you see it: the system generates scalable cognitive labor at near-zero marginal cost; the physician’s license upcycles that output into a high-value, billable medical act; the health system captures the margin; the platform provider captures the data and the standard. In genuinely high-complexity or adversarial cases — the septic presentation, the psychiatric emergency, the diagnostic puzzle — physician judgment still dominates and supervision is real. But the system is optimized for the median case, and in the median case, review collapses into ratification. That is where the economics scale. That is where formal authority and practical epistemic control diverge.
Formal authority remains with the physician. Practical epistemic control migrates to the system. And the asymmetry is invisible because it is hidden inside a relationship that is described, in all official language, as collaborative oversight. The physician becomes a liability sponge — the last mile of a new billing model that has been architecturally designed to look like a safety mechanism.
Remove the signature requirement and the system is a tool. Keep it and you have monetized clinical authority as a service. The supervision clause is not what makes it safe. It is what makes it profitable.
The shortfall is not context. It is the operating condition that forecloses competition.
The 10 million worker shortfall is real. This matters to the argument, not as a complication but as its sharpest edge. The humanitarian cover story works not because it is false, but because it is true enough that opposing it sounds callous. You cannot argue against augmentation when the alternative is no care at all. That is not a rhetorical trap. That is the actual clinical situation in underserved geographies, in low-income healthcare systems, in the jurisdictions that will be the first deployment targets because their regulatory thresholds are lowest and their alternatives are fewest.
The structural consequence of a genuine shortage is that it eliminates the competitive alternative to the system being deployed into it. There is no human labor waiting to fill the gap if the system underperforms. There is no redundant capacity to fall back on if the simulation-deployment gap is larger than the validation theater implied. Slower, human-intensive care models simply cannot clear demand at politically acceptable levels. The shortage does not just justify rapid adoption — it creates the conditions under which rapid adoption cannot be reversed, because reversal means returning to a care deficit that the healthcare system cannot politically or practically sustain.
This is why the workforce displacement analysis produces a different result depending on which population you measure. Healthcare systems facing the shortage benefit: they get throughput at reduced cost. Non-physician healthcare workers — nurses, physician assistants, community health workers — are the victims: their roles are encircled by a system that performs their functions under physician supervision at a fraction of the labor cost. The humanitarian framing makes the first group visible and the second group invisible. That invisibility is not incidental. It is structurally produced by the framing itself.
Who this is actually addressed to
The real audience for this analysis is not clinicians, who will experience the lock-in from inside it and may not recognize it as such until it is complete. It is not patients, who will experience it as the way care simply works now. It is liability insurers, health system general counsel, and the medical board members who will be asked, in the next 18 to 24 months, to treat validation studies as evidence of standard practice.
Because those actors — not regulators, not platform providers, not academic medical centers — are the ones who will inadvertently seal the architecture. When a malpractice carrier begins pricing system non-use as an elevated risk factor, the decision has been made. When a health system GC advises that documented AI consultation is now the defensible posture, the standard has moved. When medical boards begin codifying AI utilization into continuing education requirements and documentation standards, what was once a risk response has formalized into obligation. They will not be making policy decisions. They will be responding rationally to an incentive structure that has already been designed around them — and in doing so, they will have made the policy decision that no one explicitly chose.
This is what makes the capture structural rather than conspiratorial. No single actor needs to intend the outcome. Each actor needs only to respond rationally to the incentives in their local environment. The liability insurer prices risk. The GC minimizes exposure. The hospital follows guidance. The physician documents compliance. The standard hardens. The architecture locks. And at the end of that chain, the question of whether triadic care was good medicine has become entirely academic — because it is now not merely how medicine is practiced, but what counts as medicine at all.
cafebedouin.org 