Nolla Health, Heidi and OpenEvidence overlap in their use of AI, but their most relevant comparison is not model intelligence. It is the job each service undertakes. A patient seeking treatment, a clinician checking evidence and a practice completing follow-up work need different outputs, permissions and routes to human help.
As checked on 5 October 2026, Nolla presents a patient-facing care service, Heidi describes a clinician's care partner spanning documentation and broader workflows, and OpenEvidence's clinician-facing material centres on evidence-supported answers at the point of care. These are starting positions, not permanent or mutually exclusive categories. Nolla, Heidi's September announcement and OpenEvidence's educational material describe those respective roles.
This comparison is published by iatroX and includes iatroX among the tools discussed. The purpose is to match a service to a task, not to declare our platform or another company a universal winner. No head-to-head clinical-performance test was conducted for this article.
Start with the person asking for help
A fictional patient with a new skin concern wants to know whether they need treatment and how to obtain appropriate care. A polished explanation may help, but it does not by itself resolve access, suitability, prescribing or follow-up.
A fictional GP reviewing that patient's treatment history has another need: check the relevant evidence, identify missing information and decide whether the existing plan remains appropriate. The GP is not necessarily looking for another service to take over the patient's care.
An administrator arranging an already agreed follow-up has a third task. They need the correct action, recipient and confirmation of completion. They should not have to infer a new clinical plan from a generated paragraph.
These distinctions prevent an apparently impressive feature from answering the wrong question. A service can be good at evidence retrieval without operating a care pathway. A service can provide care without being the right environment for postgraduate examination preparation.
Map the tools onto the encounter
The following is a documented-role comparison as of 5 October 2026. "Not established" means the sources reviewed do not establish that capability for the particular product or deployment; it is not a claim that the vendor could never offer it.
| Platform and user | Intake and clinical context | Evidence and documentation | Treatment, prescribing and follow-up | Professional learning |
|---|---|---|---|---|
| Nolla Health: patient seeking care | Patient questions and images within the relevant service | Assessment and treatment information; not assessed here as a clinician literature-search replacement | Patient-facing care; a limited Utah acne-prescribing pilot; scope and review depend on the service and phase | No clinician examination-preparation role established in the reviewed material |
| Heidi: clinician or clinical team | Consultation context and, for announced workflows, connected practice information | Scribing, evidence access and preparation of clinical work | Announced agents support specified administrative actions with workflow-dependent approvals; unrestricted independent prescribing authority is not established | Clinical information may support learning, but an examination curriculum should not be inferred |
| OpenEvidence: clinician-facing evidence experience | Clinician supplies the question and relevant context | Citation-backed answers and evidence exploration | An evidence answer is not itself an issued prescription; an autonomous prescribing pathway is not established by the sources used here | Educational resources support use of the evidence experience; other learning functions require separate verification |
| iatroX: clinician or medical learner | A reference question, attempted examination question or practice case | Ask-iatroX provides source-linked reference information; Tutor develops the learner's reasoning | Not a patient-treatment or prescribing service; no practice-inbox execution is claimed | Question banks, Socratic Tutor, study planner, simulations and CPD tools |
The Nolla and Heidi rows follow their dated product descriptions above. The OpenEvidence row concerns its clinician-facing positioning, not a claim that the company has no other patient-facing products. The iatroX row reflects its September 2026 product information and published methodology.
Generating text is not executing an action
A useful assessment separates four events. The system suggests that a treatment might be relevant. It drafts information for a clinician. It prepares an order within an authorised workflow. It issues a prescription through a legally permitted care arrangement. Those events can occur in one service, but evidence for the first does not establish permission for the last.
Nolla's 5 October announcement concerns a restricted route to initial acne treatment in Utah. The state register describes topical treatment for mild to moderate acne in adults and staged physician oversight. It does not authorise every possible use of the Nolla app or establish nationwide autonomous prescribing. Utah's authorised-pilot register.
Similarly, an agent's ability to operate a browser or follow a practice process does not establish the authority to make a new treatment decision. The purchaser must inspect the actual action, the accountable service and the point at which approval is required. A confirmation screen is useful only when its meaning is clear.
For evidence tools, the equivalent boundary is between finding support for an option and deciding that the option applies to this patient. Sources can inform that decision without taking responsibility for obtaining missing history, interpreting an examination or discussing the patient's preferences.
Heidi is not just a scribe, but availability still matters
Heidi's announcement on 28 September 2026 described agents, memory and research access alongside its existing documentation-related products. It said rollout would begin on 29 September, with approval controls depending on the workflow and organisational settings. The announcement explicitly excluded the new capabilities from the UK and EU. That exclusion should not be extended to all existing Heidi services. Heidi II launch release.
A UK comparison therefore needs two columns in the buyer's mind: what has been announced internationally and what the organisation can actually obtain and use. Treating an international demonstration as a locally available feature can distort a purchasing decision before any clinical evaluation begins.
Enterprise configuration also matters. A documented capability does not establish that every customer receives identical record access, connectors or permissions. The relevant demonstration is the proposed workflow in the intended setting, including what happens when information is missing or the action cannot be completed.
Follow one hypothetical encounter without inventing integrations
Suppose a patient brings a treatment summary from an external digital dermatology service to their usual clinician. The clinician first establishes what happened: the presenting concern, treatment started, adverse effects, outstanding uncertainty and follow-up arrangements. That record should be evaluated on its content, not accepted or rejected because AI contributed to it.
The clinician might then consult an evidence tool about a general management question. They may separately use an approved documentation assistant to prepare the consultation record. Later, the clinical uncertainty could become a learning question explored through a question bank or Tutor.
This is an illustrative division of work, not a claim that Nolla, Heidi, OpenEvidence and iatroX have a shared integration. Moving patient information between services requires an appropriate, approved arrangement. For UK clinicians, the GMC's remote-care principles, checked on 5 October 2026, remain relevant to adequate information, consultation suitability and continuity.
The same example shows why buying overlapping tools is not automatically wasteful. Two products can complement each other when one completes an operational task and another supports understanding. Conversely, two subscriptions producing similar explanations may add little unless the second changes the quality or convenience of the work.
Where iatroX fits, and what its price actually buys
Ask-iatroX and free question access are genuinely free, without a trial expiry or verification gate. In iatroX's September 2026 product information, the paid subscription combines question banks, Socratic Tutor, the study planner, simulations and CPD tools. Simulations and CPD are included, not separate add-ons.
In iatroX's September 2026 UK pricing, the subscription is £99 a year paid upfront, equivalent to £8.25 a month when billed annually, or £29 a month. At those prices, three monthly payments total £87 and four total £116, so annual payment becomes cheaper from the fourth month. This is a professional-learning purchase, not a price comparison with patient consultations, medicines or a US treatment subscription.
The relevant value question is whether several learning methods serve the same professional goal. A learner might attempt a question, explore a misconception with Tutor, practise applying the principle and review a personal learning record. Neither breadth of unrelated examinations nor a subscription price establishes better patient outcomes.
The verdict depends on the task
For an eligible patient seeking a defined care pathway, Nolla belongs in an assessment of treatment services, including access to clinicians and follow-up. For a clinician or team trying to reduce documentation and related administrative work, Heidi's locally available and approved workflows are the relevant subject of evaluation.
For a clinician seeking literature-supported answers, OpenEvidence's evidence experience is a more direct comparison than an acne-prescribing app. For UK-oriented clinical reference and structured professional learning, iatroX offers a different combination of source checking and practice. These choices should be evaluated against the same underlying standard: does the tool help the intended user complete the right task without obscuring uncertainty or responsibility?
Frequently asked questions
Is Nolla Health an alternative to OpenEvidence?
Not for the same primary task: Nolla's patient-facing care pathway and OpenEvidence's clinician-facing evidence experience address different needs. The correct comparison depends on whether the user needs treatment delivery or information to inform a clinical decision.
Is Heidi only a documentation tool?
No, Heidi's September 2026 announcement describes evidence and broader agent-supported workflows alongside documentation. Availability and approval arrangements depend on the product, region and deployment.
Can clinicians use these categories of tools together?
Yes, a care service, an approved workflow tool and a reference or learning platform can serve complementary purposes. Their combination still requires clear responsibilities and appropriate information-sharing arrangements, rather than an assumption of integration.
Check clinical questions against linked sources with Ask-iatroX →