Google’s entry makes clinical AI more competitive. It also makes the choice more specific. A hospital looking to coordinate systems, a physician investigating evidence and a trainee trying to improve their reasoning are buying different kinds of help. Comparing every product as a medical chatbot obscures those differences.
Google launched Gemini for healthcare in preview today, aimed at connected clinical and administrative workflows. Google Cloud’s introduction
The useful comparison with OpenEvidence and iatroX begins with the work surrounding an answer. How does the system acquire context? What sources govern its response? Where does the clinician review it? And what happens when the answer reveals a gap in the clinician’s own understanding?
For iatroX, that last question is a particularly strong place to compete. Its case rests on clinical knowledge and professional development that clinicians can use deliberately, across the settings in which they practise and train.
Three starting points for the same clinician
Consider a doctor preparing for an unfamiliar clinic. There may be three separate needs: understanding a patient’s fragmented history, investigating a disputed clinical question and rehearsing the decisions the clinic will require. A product can help with more than one, but success on one does not establish success on the others.
An enterprise workflow system earns its place by fitting organisational data, access and processes. An evidence assistant earns its place by making a clinical question easier to investigate and verify. A learning platform earns its place by helping the doctor identify and work through a weakness, then return to it.
That distinction gives a buyer a better starting point than asking which AI is most intelligent. Intelligence matters, but the product must make that capability useful within a particular responsibility.
The patient-history problem is about an institution’s systems. The disputed-evidence problem is about the quality and applicability of information. The learning problem is about the clinician’s development. These responsibilities overlap without becoming interchangeable.
OpenEvidence has moved well beyond basic search
OpenEvidence should be assessed as it exists now. Its September model launch introduced Osler for faster answers, Sackett for deeper investigation and Snow for extended research; Snow succeeds DeepConsult. Darwin is a separate application-only research preview. Those choices let eligible users trade response time for investigation depth, although the company’s benchmark results do not establish superiority in every clinical setting. OpenEvidence’s model announcement
Its institutional offering also changes the comparison. OpenEvidence Enterprise advertises SSO, EHR launch, contextual evidence where supported, institutional agreements, retention controls and no model training on sensitive clinical data. It is offered at no cost to eligible provider organisations supporting verified US clinicians. Integration routes and context availability depend on the organisation. OpenEvidence Enterprise
The announced NYU Langone collaboration places patient-aware evidence answers inside Epic. That is evidence of a broader product direction; it is not evidence that every organisation already has the same deployment. OpenEvidence and NYU Langone’s announcement
This matters competitively. A familiar tool embedded in a clinician’s working environment can be easier to reach than an excellent tool elsewhere. The clinician opening iatroX for a revision session or to investigate an error in their reasoning is expressing a different intention from a clinician checking a chart. Each experience can earn its place by serving that intention well.
A referenced answer still needs the right clinical context
The presence of citations is a beginning. A clinician must also consider whether the source addresses the population, intervention, decision and practice setting in question. A study can be accurately represented yet provide an incomplete basis for the next local action.
That is why guidance and research should be distinguished. Research can establish benefits, risks or uncertainty. A local guideline can incorporate those findings into a recommendation shaped by available services and an accountable process. Both are valuable; their roles differ.
iatroX’s published standards describe accepted guidance as the starting point, with research used to add context. They also distinguish UK clinical guidance-led use from international education and exam use. This is a useful statement of product scope, rather than proof of comparative accuracy. iatroX’s clinical AI standards
For a clinician, the practical advantage is a clearer question to ask of the tool: which context is governing this response? For an educator, it is whether the answer is helping someone learn the decision they will actually be expected to make.
A platform deserves credit when it makes that distinction visible. It should not require users to infer jurisdiction from the spelling of a medicine or the organisation mentioned in a citation.
Geography means access and applicability
These are separate checks. OpenEvidence’s published standard terms, updated on 17 September, prohibit access from the UK, EU and EEA; its acceptable-use policy separately restricts inputs concerning residents of those regions. Those terms should be read alongside any applicable institutional agreement. They do not justify calling the entire OpenEvidence business US-only. OpenEvidence’s terms, acceptable-use policy
The American Academy of Ophthalmology's guide lists access for ophthalmologists in countries including Australia and Canada, as of 22 June 2026. This reinforces the need to distinguish a particular UK restriction from a claim about worldwide access. AAO OpenEvidence access guide
iatroX’s opportunity extends beyond serving a geographic gap. A learner in a market where OpenEvidence is available can still want structured preparation for a specific examination, a question-centred learning conversation or a professional record of what changed. Those are positive reasons to choose a product.
Learn makes education a serious comparison
OpenEvidence Learn launched on 4 November. Its offer includes more than 50,000 board-style questions, simulated-patient tutoring, flashcards and CME, free to OpenEvidence users. That makes a comparison limited to clinical search versus revision outdated. OpenEvidence Learn’s announcement
The next comparison must examine educational fit. Does the relevant bank match the examination? Does feedback address the learner’s chosen answer? Can someone distinguish a factual gap from a recurring reasoning error? Is the next practice activity appropriate to the weakness just exposed?
Free Ask iatroX sits alongside exam preparation spanning UK and international pathways. That breadth matters when a career crosses jurisdictions or moves from licensing into specialty training. iatroX’s platform overview
Its Tutor works beside a question the learner has attempted, offering guidance or direct explanation and helping retain an insight as a highlight or editable Anki card. iatroX Tutor
The value of that arrangement is continuity. The learner does not have to reconstruct the question, their choice and the relevant distinction before beginning a useful conversation. It creates a concrete place to examine why a plausible answer was weaker than another.
Established reference products remain important
The market also includes providers with a long-standing editorial foundation. UpToDate Expert AI adds conversational access to its clinical reference approach. Wolters Kluwer’s August update describes enterprise adoption, international pilots and medication guidance with supporting assumptions; regional rollout and individual capabilities still require checking. Wolters Kluwer’s update
Dyna AI Mode operates within DynaMed, DynaMedex and Dynamic Health, allowing clinicians to move between AI synthesis and traditional exploration of expert-curated material. EBSCO’s February announcement specifies an EU availability restriction, which should not automatically be treated as a UK restriction. EBSCO’s announcement
These alternatives make evidence provenance a competitive necessity. A specialist platform must offer more than citations attached to fluent text. Its source choices, clinical scope and review experience must help users understand what the answer can support.
Learning should leave something useful behind
iatroX’s available simulations provide voice or text practice where supported, followed by a transcript, evidence-linked feedback and Tutor follow-up. iatroX’s access and simulation details
Simulation introduces a different test: can the learner use what they know while an interaction unfolds? It can expose timing, prioritisation and communication gaps that a correct written response leaves hidden. An educational encounter remains a rehearsal, so its value should be assessed through the specific practice and feedback it provides.
iatroX’s CPD workflow connects targeted activity, Tutor-supported learning and a reflection the user reviews before confirming an exportable record. iatroX CPD
That gives the clinician a purposeful destination for a discovered weakness. The aim is to make the learning visible and revisitable. A completed record matters because it describes an activity and a personal next step, rather than merely documenting that an answer appeared on a screen.
Where iatroX makes the strongest case
An organisation may need an enterprise agent. A clinician may value OpenEvidence or an established reference service for evidence investigation. The same clinician can have a compelling reason to choose iatroX for deliberate, context-specific development.
The strongest comparison follows a question through its consequences: find the relevant information, identify what remains uncertain, test understanding, examine the reasoning and practise using it. iatroX places those professional needs close together.
That is a substantial role in the market Google has entered. As AI makes answers easier to obtain, a product that helps clinicians become better at understanding, questioning and applying them has a clear proposition of its own.