Research · November 1, 2026

Vera Health's $28.5M Series A — Grading the Evidence Before Answering the Doctor

Vera Health raised a $28.5 million Series A led by Threshold Ventures to grade clinical evidence before answering doctors' questions.

What happened

On October 27, 2026, Vera Health, a San Francisco-based clinical AI search platform, announced a $28.5 million Series A led by Threshold Ventures, with participation from Mantis VC and SV Angel. Existing investors Y Combinator and Gradient increased their stakes — the company says they "tripled down" — bringing total funding to $32 million. The announcement came on the company's own blog, not through a wire release, and the figures below are company-reported unless stated otherwise.

Vera Health answers clinicians' questions from medical literature and guidelines. It is not a general-purpose chatbot with a medical prompt on top. The product claim is a reversal of the usual order: find the evidence first, then write the answer. It searches more than 60 million peer-reviewed publications, clinical guidelines, consensus statements, and drug labels, links every claim to its supporting research, and color-codes the strength of the available evidence.

The red grade is the product

"Chatbots are built to sound confident. Vera is built to be accurate, and honest when the evidence is thin," the announcement says. That sentence is the whole thesis, and the mechanism that makes it concrete is the grading itself.

A rare condition may have only a small case series or a single-patient report behind it. Vera surfaces that research while marking its limitations visible — sometimes, the company says, in red — so a physician can judge how much weight the finding deserves. Co-founder Maxime Allouch frames it as a retrieval problem as much as a writing problem: "The grade shapes what Vera retrieves and writes, while the clinician brings the patient, the context and the judgment."

That distinction matters because the failure mode of medical AI is not usually a wrong answer stated hesitantly. It is an uncertain answer stated confidently. A system that blends a large randomized trial and a case report into the same assured paragraph destroys exactly the information a clinician needs most: how much to trust each source. Vera's bet is that making uncertainty visible is not a weakness to hide but the feature clinicians will trust — and, crucially, the feature that makes the tool defensible against general-purpose models that can already answer medical questions fluently.

The company pairs the grading claim with an institutional one. Through a partnership with the American College of Emergency Physicians (ACEP), the platform integrates ACEP clinical policies with source attribution. Scott M. Silvers, MD, chair of ACEP's Clinical Policies Committee, is quoted in the announcement: "A large, randomized trial and a case report should never carry the same weight, yet most AI tools blend them into the same confident answer. Level of evidence is the foundation of every guideline we write, and Vera Health is built on it."

Distribution without a sales force

The growth numbers, all company-reported, are striking precisely because of the team size behind them. A year ago, a few thousand clinicians were using Vera Health. Today, the company says, more than 800,000 are, most having discovered the platform through a colleague. With a team of just 12 employees, it says it has reached clinicians in 190 countries, including approximately 1 in 4 physicians in Spain and Italy and 1 in 7 in France, largely through word of mouth. Every month since June, it says, Vera has been the most downloaded clinical app in the world across the App Store and Google Play — a ranking the company attributes to Appark download data for June through September 2026.

Two cautions belong next to those numbers. First, "users" and "downloaded" are not "paying customers" or "daily active clinicians": the announcement gives no paying-versus-free split, no retention figure, and no revenue number, so the adoption curve cannot yet be read as a business curve. Second, the download ranking is company-cited third-party data, not independently verified here; Appark's underlying data was not reviewed.

Still, the shape of the distribution is informative. Medicine is a referral profession: a tool a doctor trusts on one hard case gets recommended in the corridor, not through an advertising funnel. Co-founder Taieb Bennani makes the point directly: "Everyone on our team is a clinician, or the child or sibling of one, like Maxime and me." For a product whose adoption depends on professional trust, a team built from inside the profession is a distribution claim as much as a hiring detail.

The announcement also offers one clinical anecdote, told by Dr. Bernard P. Chang, Vera Health's Chief Scientific Advisor, about an emergency-department patient whose cough and shortness of breath could have been read as a routine respiratory illness; Vera surfaced evidence that raised concern for pulmonary embolism, a potentially life-threatening blood clot in the lungs, and led him to broaden the evaluation. That is a single case, told by an advisor to the company, not a controlled accuracy study — and it should be read as an illustration of the intended use (surfacing evidence that challenges an initial assumption), not as proof of diagnostic performance.

What the money is for — and what it is not for

The financing will support further development of Vera Health's clinical evidence infrastructure, expansion into additional healthcare markets, and deeper partnerships with medical societies, publishers, and health systems. The company says Vera Health will remain free for licensed clinicians and medical trainees worldwide.

That last sentence defines the business-model question the round leaves open. Free for clinicians, paid for by whom? The plausible answer is institutions — hospitals, health systems, and medical societies buying enterprise access — but the announcement names no enterprise pricing, no paying institutional customer count, and no revenue figure. The Series A buys time to convert a free, trusted, widely distributed clinician tool into an institutional sale. Whether that conversion works is the experiment the funding funds.

The mission framing in the announcement points at the size of the gap rather than the size of the market: new medical knowledge takes an estimated 17 years to reach routine clinical practice, according to research in the Journal of the Royal Society of Medicine cited by the company. "Our mission is to close the gap between what medicine knows and what every patient receives," Bennani said. "Today, that gap is measured in years. We want it measured in seconds."

What this does not show

The announcement discloses no valuation, no revenue, no pricing, and no accuracy benchmark against competing clinical tools or general-purpose models. The 800,000-clinician figure is a registered-user-scale claim with no activity, retention, or paying-share breakdown. The most-downloaded-app ranking rests on Appark data the company cites but does not publish. The ACEP partnership is described as policy integration with source attribution, not as an endorsement of any particular answer's correctness. And the pulmonary-embolism anecdote, however vivid, is one case, not evidence of systematic improvement in clinical reasoning.

The competitive question is also open. Incumbent reference tools and general-purpose AI systems already answer clinical questions; Vera's argument is that none of them grades evidence the way a guideline committee does. If that grading proves accurate, auditable, and fast enough to use on shift, it is a genuine moat — evidence quality is hard to fake and slow to build. If clinicians treat the grade as decoration and read only the answer, it is a feature, not a moat.

What would turn this round from a thesis into evidence is the institutional version of the numbers: how many health systems pay, what share of the 800,000 clinicians are active weekly, how often the red grade changes a clinical decision, and whether graded answers outperform ungraded ones on a pre-registered accuracy test inside the academic medical centers where, as Chang puts it, medicine is actually practiced.

Sources