Doximity deepens Ask with UpToDate as clinical AI competition shifts toward trust
Doximity is folding UpToDate directly into Ask, a move designed to make the company’s clinical AI tool feel less like a generic chatbot and more like a physician-facing reference desk built into workflow. In the coming weeks, clinicians at health systems that use both products will see an answer based on UpToDate’s evidence-based clinical content alongside Ask’s response, without leaving the interface.
The integration matters because clinical AI is no longer competing only on speed or fluency. It is competing on whether a physician can trust the answer enough to use it at the bedside. By placing UpToDate’s 13,000 clinical topics and best practice guides inside Ask, Doximity is trying to wrap a familiar evidence layer around a conversational tool that already lives inside a widely used professional platform.
A trust layer built around evidence, not just generation
Doximity’s pitch is straightforward: if a clinician is going to use AI for patient care decisions, the system should surface the underlying medical authority in the same moment the question is asked. UpToDate brings 30 years of market presence and a large editorial structure built from 7,600 expert clinicians, editors and peer reviewers who regularly review clinical, drug and patient content. Doximity, in turn, brings distribution. Its membership includes more than 85% of all U.S. doctors, and its AI tools have been deployed across more than 165 health systems, including 10 of the nation’s top 20 hospitals.
The move also extends Doximity’s existing strategy of using recognized clinical resources to strengthen confidence in its own AI products. Ask now sits alongside the company’s 12,000-physician PeerCheck program and its broader suite of workflow tools, including Scribe and the clinician-to-patient Dialer. In practical terms, Doximity is trying to make its platform the place where doctors ask, verify and act, rather than a stopover on the way to other clinical references.
Jeff Tangney, Doximity’s chief executive, framed the partnership as a way to answer the questions that published trials and journals do not always settle. A substantial share of medical questions still has no direct answer in the available evidence, and that gap is exactly where real-world clinical judgment matters most. UpToDate has long occupied that space by combining evidence review with expert interpretation, which is why the pairing with Ask is aimed less at replacing a clinician than at shortening the path to a defensible answer.
One sentence captures the business logic cleanly: Doximity is making evidence the interface.
Why the clinical workflow argument is the real product story
The technical novelty here is not that AI can generate an answer. Large language models can do that quickly, and increasingly well. The harder problem in medicine is reliability under uncertainty, especially when the literature is thin, conflicting or silent on a corner case. Tangney said every frontier model has read the same journals, but UpToDate stands apart because it reflects experienced physician judgment in situations where there is no randomized controlled trial that settles the question.
That distinction has direct bedside implications. A clinician searching for the next step in a complex patient with multimorbidity, an uncommon medication interaction or an edge-case diagnostic dilemma often needs more than a synthetic summary of published papers. The answer must be filtered through expert interpretation, not merely extracted from text. Embedding UpToDate inside Ask gives Doximity a way to present that expert layer without making the doctor leave the workflow, re-enter credentials or triangulate among multiple browser tabs.
Nicholas Perkins, DO, chief clinical analytics officer at Prisma Health, put the use case in concrete terms: clinicians depend on trusted resources like UpToDate when evidence is scarce, and immediate access inside Doximity Ask may save time that can be redirected to patient care. The clinical significance lies less in novelty than in friction reduction. When a tool sits inside the point-of-care routine, the odds rise that it will be used in the moment decisions are being made.
Generative AI is being pulled toward curation and citation
Doximity is not alone in moving clinical AI toward a more curated model. Wolters Kluwer launched UpToDate Expert AI last year, bringing generative AI into its own clinical decision support ecosystem. It has also partnered with Epic and Abridge, while Suki teamed with Wolters Kluwer to launch Ask, powered by the UpToDate Expert AI API. The broader market signal is clear: health-tech companies are learning that generative performance alone is not enough when physicians are evaluating whether to rely on an answer in a clinical setting.
Trust issues remain central. Tangney said trust is the number one issue with AI answers, and that assessment tracks with the pressure hospitals face as they evaluate enterprise AI platforms. The clinical stakes are different from consumer search or routine productivity software. An inaccurate recommendation can alter prescribing, referral patterns or diagnostic sequencing. A system that cites a respected evidence base can narrow the gap between convenience and confidence, but it does not eliminate the need for physician judgment.
Doximity is also trying to distinguish itself through physician-focused validation. Tangney pointed to ARISE benchmark research led by physicians from Stanford University School of Medicine and Harvard Medical School, which found that clinically specialized AI tools, including Doximity Ask, outperformed general-purpose models in simulated cases. He said the field needs more rigorous, independent physician-led research of that kind. The message is less about hype than about comparative safety: specialized tools should be measured against the real risks clinicians face, not against abstract language benchmarks.
Doximity is building a broader physician operating system
The UpToDate deal fits into a wider Doximity strategy that reaches beyond decision support. The company wants Ask, Scribe and Dialer to function as a single digital assistant for physicians, one that covers communication, documentation and clinical lookup within one system. Tangney described those tools as the physician’s personal assistant, even likening the bundle to “the Microsoft Office of being a physician.” The metaphor is commercial, but the clinical idea is tangible: if a physician can speak to patients, document care and consult evidence in one environment, the workflow becomes less fragmented.
Doximity’s AI usage metrics suggest that physicians are adopting the tools quickly. Quarterly active workflow prescribers grew more than 30% year over year to record highs, nearly half of those prescribers used the company’s AI tools in the first quarter, prompt volume rose more than 25% quarter on quarter and Scribe note-taking users increased 10-fold from July 2025 to July 2026. Those figures do not prove clinical superiority, but they do show that physicians are willing to fold AI into daily practice when the product is embedded in familiar routines.
The next frontier is obvious, even if Doximity is not spelling it out. Ask currently focuses on helping clinicians make decisions, and a logical extension would be tools that help carry those decisions into action, such as prescribing support or referral drafting. Tangney declined to provide roadmap details, leaving the near-term story centered on distribution, trust and evidence curation rather than deeper automation.
For now, the UpToDate integration is a statement about how clinical AI will be judged. Speed matters, but authority matters more. In a market crowded with models that can generate plausible language, the tools most likely to gain durable clinical use will be the ones that can show their work, name their evidence and stay inside the physician’s existing workflow.