Heidi says it reached US$50 million in annual recurring revenue in April 2026. That is a dated company-reported milestone, not an audited statement of revenue earned during a completed year and not a current revenue result measured on 22 September 2026. It can inform discussion of willingness to pay, but it cannot establish the size or durability of a clinical benefit. Heidi's September release is the source cited for the milestone.
This article is published by iatroX and includes iatroX's own reach figures to apply the same reporting discipline. It does not use private financial information or estimate a competitor's missing revenue.
What ARR is trying to describe
Annual recurring revenue is a recurring-revenue run-rate measure. It attempts to express recurring business on an annualised basis, but the precise calculation depends on what the company includes and excludes. It is not interchangeable with revenue recognised in financial statements, cash received or profit.
A contract can contribute to a recurring-revenue measure while payments arrive on a different schedule. A company can also generate substantial recurring revenue while incurring substantial costs. The headline alone does not settle either issue.
For Heidi's reported April 2026 milestone, the appropriate description is therefore "company-reported ARR at that date". Unless the source discloses the calculation and supporting accounts, the reader should not silently add the words "audited", "cash revenue" or "profitable". The company release provides the reported figure, not those additional conclusions.
Preserve the date as carefully as the amount
The milestone belongs to April 2026 even though it appears in September funding coverage. It should not be presented as a new September measurement, nor should an analyst assume that it remained unchanged or grew at a particular rate afterwards.
Heidi's financing announcement on 22 September 2026 separately describes a US$100 million Series C and US$240 million growth investment. Funding and ARR are different categories: one concerns financing, while the other concerns recurring commercial activity. Adding them together would produce a number with no useful business meaning. Heidi's financing announcement supplies the transaction figures.
The disciplined approach is to preserve the metric, currency, measurement date and attribution together whenever the milestone is repeated. That gives readers a clear boundary between reported data and later interpretation.
The questions behind commercial traction
The next question is who is paying and why. An individual subscription, a departmental deployment and a large institutional contract can represent different purchasing behaviour. Without a disclosed breakdown, an analyst should not assume that one model dominates.
Retention is another question. Do customers renew because the service remains useful? Do they expand into additional roles or sites? Does a deployment need substantial continuing support? The ARR headline cannot answer those questions, but they influence how commercial traction should be interpreted.
Customer acquisition also matters. General Catalyst's published Customer Value model, checked on 22 September 2026, describes financing acquisition separately from equity-funded product risk. That explains why growth economics deserve attention in the Heidi story, without revealing the economics of Heidi's own agreement. General Catalyst's explanation provides the general financing context.
A fictional investor's measurement problem
Imagine an analyst reviewing two clinical AI companies. One reports a recurring-revenue milestone, while the other reports the number of organisations reached. The analyst wants a single ranking.
The correct response is not to invent an average revenue per organisation for the second company or assume that the first has more active clinicians. The analyst should create separate categories: commercial run rate, organisational distribution, active use, retention and measured effects on work. Empty cells remain empty until comparable data are available.
A further complication arises if one company's organisational figure includes pilot access while another's includes only paid deployments. Even identical units can hide different definitions. A useful report preserves the original wording and asks for clarification rather than manufacturing comparability.
What can be compared with Tandem?
Tandem's 14 September 2026 funding announcement reports its financing and operational reach, but the announcement reviewed does not supply a directly comparable ARR figure. Tandem's release therefore supports discussion of its announced expansion and financing, not an invented revenue estimate.
The two companies can be compared by the categories they disclose. Their September 2026 US$100 million equity rounds are financing figures. Heidi's April 2026 ARR is a commercial run-rate figure. Organisations served, countries reached and consultations processed belong in different operational categories.
None of those measurements should be used as a substitute for an unavailable one. In particular, consultations processed should not be converted into paying users, and a geographical footprint should not be treated as evidence of a particular retention rate.
Apply the same standard to iatroX
Per iatroX platform data supplied in September 2026, cumulative reach comprises 1 million uses, over 50,000 verified clinicians across 40 countries, primarily the UK, including over 15% of UK GPs. These are founder-supplied reach figures, not audited active-user counts, paid-subscriber counts or measured learning outcomes.
The reported figures do not specify the definition of a use, the accumulation period, the exact September 2026 cut-off, the clinician-verification method or how countries are assigned. The numerator, denominator and reference date behind the UK GP percentage are also unspecified in the reported data, limiting direct comparisons with other platforms.
The cumulative figures must not be assigned to iatroX Simulations or CPD tools simply because those features are now part of the platform. Nor should a clinician count be recast as a student count. A whole-platform adoption statement does not establish adoption of a later-launched feature.
For reliability, the relevant discussion concerns sources and methodology. As described in September 2026, iatroX's published methodology includes retrieval, ranking, citation grounding, output checks and uncertainty handling. Those are design features to inspect, not proof created by a large usage count.
Commercial demand and clinical benefit can diverge
A purchaser may value reduced documentation friction, a familiar interface, integrated support or clearer access to information. Commercial demand can reveal that something is worth paying for without showing precisely which benefit drove the purchase.
Clinical or educational claims need their own evidence. A study of review time should not be retold as proof of improved patient outcomes. A satisfied learner should not be counted as an examination pass. A completed CPD record should not be treated as a validated change in practice.
The implication is not that commercial traction is unimportant. It is that commercial and clinical questions deserve their own measures. Keeping them separate makes strong results more credible and uncertainty easier to understand.
Verdict by reader scenario
For an investor, Heidi's reported April 2026 milestone is a useful starting point for questions about recurring demand and retention, not a complete valuation model. For a health-system buyer, require evidence about the intended workflow and implementation. For a clinician, evaluate the service's actual usefulness rather than its revenue. For a learning-platform reader, keep iatroX's reach separate from evidence of learning effectiveness.
Frequently asked questions
Did Heidi report US$50 million of completed-year revenue?
The reported milestone is US$50 million ARR reached in April 2026, according to Heidi's September release. It should not be relabelled as audited annual revenue or cash received.
Can Heidi and Tandem be ranked by their published adoption figures?
Not where the figures use different units, definitions or periods. Compare the categories separately and leave missing data unestimated.
Do iatroX's cumulative figures describe simulation usage?
No: the September 2026 figures concern whole-platform cumulative reach. They should not be assigned to a later-launched feature or interpreted as evidence of learning outcomes.