A free clinical AI service can be useful, expensive to operate and commercially rational at the same time. The absence of a charge to the clinician does not explain who funds the product, what the company is trying to achieve or which features an institution may still need to purchase.
The right response is not to assume that free means compromised. It is to make the economic relationship visible. A reader who understands the payer can ask better questions about access, continuity, governance and product priorities.
Doximity has more than one customer relationship
Doximity's June 2026 quarterly filing describes revenue from marketing, hiring and workflow subscriptions. Workflow offerings include its Clinical AI Suite. This makes a simple description of Doximity as either a free app or an advertising business incomplete.
Different users and buyers can value different parts of the same platform. A clinician may use a reference feature without paying personally. A health system may pay for an enterprise workflow. Another customer may purchase a separate commercial service.
The strategic question is how these activities support one another. Useful clinical tools may strengthen engagement and institutional relationships. That is a business inference; it is not evidence that a particular clinical answer has been influenced by a commercial customer.
OpenEvidence states an advertising-funded approach
OpenEvidence's advertising policy says advertising helps fund free access and describes separation between advertisements and answer content. That stated separation should be represented accurately in a comparison.
The relevant scrutiny concerns how the policy is implemented and what users can observe. Are advertisements clearly distinguished? Are clinical claims traceable to sources? Is there a process for reporting concerns? A serious assessment asks these questions rather than treating the existence of advertising as proof of manipulated recommendations.
It is equally unhelpful to assume that a published policy resolves every possible issue. Policies define commitments; governance and evaluation show how those commitments operate. The distinction applies to any vendor's assurances, regardless of business model.
Subscriptions create incentives too
A subscription-funded service has reasons to retain customers, encourage renewal and justify higher-value packages. An institutional supplier has reasons to demonstrate adoption and expand deployment. A learning platform has reasons to persuade users that continued access is worthwhile.
These incentives can align with useful products. They can also encourage exaggerated claims, unnecessary bundling or metrics that flatter engagement. No revenue model removes the need for scrutiny.
A useful comparison therefore asks what behaviour the business model rewards and what safeguards separate that behaviour from clinical content. It avoids the convenient but weak argument that one's preferred charging model guarantees neutrality.
A content integration can have several payers
The Doximity-UpToDate arrangement could involve existing customer entitlements, platform fees, usage charges, new subscriptions or a combination. Its commercial terms have not been disclosed in the announcement. The deal-structure analysis examines those possibilities separately.
For the clinician, an integrated answer might appear without an additional personal charge even when money changes hands elsewhere. For the institution, the cost could sit in a reference contract, a workflow contract or a separately negotiated extension.
This is why "free UpToDate inside Ask" would be an imprecise description unless the entitlement and payer are actually established. Convenience at the point of use is not the same as a zero-cost supply chain.
The cost that does not appear on an invoice
Time is part of the practical cost. A service that generates an answer quickly but requires repeated searching to verify it may impose more work than a slower first response with clear supporting passages.
Attention also matters. Notifications, competing features and difficult navigation can make a nominally free product costly to use. Conversely, a well-integrated service can be valuable because it removes small obstacles that recur throughout a working day.
A third cost is dependence. When a clinician or organisation builds a workflow around a particular interface, changes in access, price or availability can require retraining and replacement. That does not make integration undesirable, but it should be included in a realistic assessment.
How to compare value without inventing a return on investment
Begin with a defined task and a baseline. Record the effort required to complete it before and after the proposed change, including source checking and any corrective work. Separate measured time from users' impressions of speed.
Then distinguish time released from money saved. A few minutes gained may improve the clinician's day without reducing staffing expenditure. It may create capacity that the organisation cannot immediately convert into additional activity. Both can still be worthwhile outcomes.
This distinction is important when evaluating enterprise claims. A calculation that multiplies every theoretical minute by a salary rate can produce a striking figure while overstating cash savings. A useful business case identifies how the benefit would actually be realised.
Where iatroX's model can be assessed directly
iatroX's CPD offering sits within its all-access learning subscription. That gives users a concrete bundle to assess: whether the learning activities and evidence workflow solve a need worth paying for.
The positive case is not that a subscription eliminates all incentives. It is that the user can connect the purchase to a specific activity, such as structured practice or a reviewed learning record. The same standard should govern any comparison with a free reference service.
A clinician may reasonably combine institutionally funded reference access with a personally chosen learning tool. The important question is what the additional purchase adds, not whether all products in the comparison share the same payer.
Frequently asked questions
Does advertising prove a clinical AI answer is biased?
No. It creates questions about incentives and governance, but an allegation about a particular answer requires evidence. Published separation policies should be acknowledged and their implementation assessed fairly.
Can a product be free to clinicians but paid for by hospitals?
Yes. Individual access and enterprise arrangements can coexist. The exact capabilities, support and governance included in each arrangement need to be checked rather than assumed identical.
What is the best way to decide whether a paid learning tool is worthwhile?
Identify the learning task that remains unresolved by existing resources, then assess whether the paid activity addresses it. The relevant value is improved practice and understanding, not simply a longer feature list.
Explore iatroX CPD to assess its question-led learning and evidence workflow.