A fictional patient asks to continue a medicine prescribed by their usual clinician. Another seeks their first prescription for acne. Both encounters may end with something sent to a pharmacy, but the decisions are not equivalent: one tests whether an established plan should continue, while the other must justify creating a treatment plan.
That is the central distinction between the Nolla Health and Doctronic Utah pilots described in public sources on 5 October 2026. It is a more useful comparison than asking which company has the better "AI doctor". Utah's authorised-pilot register.
This article is published by iatroX and includes iatroX's reference and learning tools in the discussion. It is a comparison of documented purposes and evaluation questions, not a hands-on test or an assertion that iatroX provides a competing prescribing service.
Continuing a plan versus creating one
Utah describes Doctronic's pilot as processing renewals of medicines already prescribed by a licensed provider, within an agreed list. It does not authorise a new treatment plan. Nolla's announced acne pilot, by contrast, permits a restricted initial-treatment pathway. The state's Doctronic account and Nolla's October announcement.
Continuing treatment is not merely administrative. The fact that a prescription was appropriate previously does not establish that it remains appropriate. A review still has to recognise information that might make continuation unsafe or require a different clinician-led decision.
Starting treatment carries an additional question: why this treatment for this presentation? That requires a sufficiently supported assessment, an eligible patient and a suitable option. A restricted choice set can make this problem more bounded, but it cannot make the question disappear.
The earlier iatroX article on Doctronic's Utah pilot and UK implications provides historical context. The Nolla development extends that discussion without converting a renewal service into an initial-prescribing service retrospectively.
Current oversight is not the same as eventual permission
There is an important update to simplified accounts of Doctronic. When checked on 5 October 2026, Utah's register said the pilot remained in Phase 1, with a licensed practitioner authorising every request. It said progression had not been approved. This should take precedence over interpreting an earlier headline as proof that autonomous renewal is already occurring throughout the pilot. Utah's current status statement.
For Nolla, the documented sequence begins with approval before prescribing and may progress to retrospective review and sampling, subject to state authorisation. Its launch announcement is not evidence that it has already reached the less intensive stages. Nolla's staged description.
A fair comparison therefore asks both "What might the agreement permit?" and "What stage is operating now?" Combining those answers into a single checkbox labelled autonomous would conceal clinically relevant information.
The information each decision needs
For the fictional renewal patient, a proposed assessment framework would establish identity, the existing prescription, the reason for treatment, relevant monitoring, adherence and material changes since the original decision. A newly reported problem should not be dismissed simply because the patient selected a refill option.
For the fictional acne patient, the framework would also consider whether the diagnosis and severity fit the pathway, previous attempts at treatment, relevant contraindications and the patient's priorities. There may be a good reason not to start a medicine even when a permitted option exists.
These are clinical-reasoning requirements, not claims that either company omits a check. Their implementation must be assessed using the actual workflow and records, rather than inferred from the brevity of a marketing page.
The distinction also helps expose a common documentation error: "not reported" is not the same as "assessed and absent". A system that accepts an incomplete answer needs an explicit next step. It should not quietly turn missing information into reassurance.
Compare safety arrangements by function
The table summarises public descriptions checked on 5 October 2026. It deliberately separates documented features from information that this review has not established.
| Question | Nolla acne pilot | Doctronic renewal pilot |
|---|---|---|
| What decision is in scope? | A restricted initial acne-treatment pathway, as well as permitted continuation | Continuation of an existing prescription within the pilot's scope |
| What is the key boundary? | Eligibility for the specified acne pathway | No new treatment plan under the renewal pilot |
| What does early oversight involve? | Physician approval before prescriptions are sent | State register says every request remains practitioner-authorised in Phase 1 |
| Can oversight change? | Later stages require state authorisation | Progression requires state approval rather than an automatic switch |
| Is clinician contact relevant? | Nolla advertises physician messaging | The pilot includes escalation to licensed care; response times were not established here |
| What records are available? | Structured intake and facial images are described; comprehensive external-record access is not established here | Identity and existing-prescription verification are described; comprehensive longitudinal-record access is not established here |
| Are there comparable independent outcome results? | Not established in the reviewed material | No matched comparison with Nolla established in the reviewed material |
Sources are Utah's register, its Doctronic page and Nolla's patient information. The table is not a safety ranking.
Why a narrower condition does not settle overall risk
There are at least two dimensions to consider: the variety of clinical situations and the consequence of the action being delegated. A service focused on one condition may have fewer treatment branches, while still making a genuinely new decision. A renewal service may continue a known treatment but encounter a wider range of reasons why continuation should pause.
Neither observation produces an overall winner. Eligibility, data quality, escalation and the consequences of errors need to be examined together. Calling something routine does not establish that every patient presenting through that route is uncomplicated.
Consider a hypothetical renewal request where the patient's account conflicts with the prescription record. The right result may be a pause and clarification. Consider an acne request where the images are clear but a history answer places the patient outside the service's scope. Again, the right result may be no automated prescription.
Those outcomes should count as appropriate decisions in an evaluation. A system should not be rewarded simply for completing a higher proportion of requests if completion includes cases that should have been diverted.
What evidence would allow a fair comparison?
Clinician agreement, treatment response and adverse events answer different questions. A reviewing physician may agree that a recommendation fits the supplied information while the service still lacks information that would change the decision. Conversely, a conservative escalation might reduce agreement on the proposed treatment while improving the handling of uncertainty.
A useful comparison would define eligible populations, observation periods, failure categories and follow-up completeness before comparing results. It would report treatment appropriateness, consequential errors, patient experience and completed escalation, with independent adjudication where feasible. It would not combine a diagnostic benchmark from one company with an operational agreement percentage from another.
No matched evaluation is reported here. iatroX's existing Doctronic evidence review discusses why the meaning of an agreement claim depends on the task and reference standard.
A verdict by reader scenario
For a patient seeking continuation of an established medicine, the renewal pathway is the relevant category, subject to eligibility and current oversight. For an eligible Utah adult seeking initial acne treatment, Nolla's restricted pathway addresses a different need. For anyone outside those scopes, neither pilot is evidence of unrestricted access to an AI prescriber.
For a clinician or learner, the useful exercise is to write down what information would make you decline either request before looking at a proposed answer. As described in iatroX's September 2026 product information, Ask-iatroX supports source checking, while its question-based Tutor explores reasoning after an attempted question. That can support learning about the distinction without claiming to validate either company's service.
Frequently asked questions
Does Doctronic issue new prescriptions under its Utah pilot?
The state's description limits this pilot to renewals of an existing prescribed treatment, not initiating a new treatment plan. Its register also described the pilot as remaining in practitioner-authorised Phase 1 when checked on 5 October 2026.
How is Nolla's pilot different?
Nolla's acne pilot includes a restricted route to an initial prescription. That changes the decision from whether to continue a plan to whether starting a treatment is justified, within the pilot's limits.
Which outcomes would allow a fair comparison?
Comparable populations, defined treatment-appropriateness measures, consequential errors, completed escalation and follow-up outcomes would be more useful than unrelated headline scores. Any comparison must also account for the different decisions and oversight stages.
Practise the reasoning behind treatment decisions with iatroX Tutor →