Mount Sinai Health System announced on March 31, 2026 that it is integrating OpenEvidence, an AI-powered clinical decision support platform, directly into its Epic electronic health record, giving every clinician across its seven-hospital New York academic network the ability to ask medical questions in plain language and get answers pulled from peer-reviewed literature without leaving the patient chart. The deal, reported by Mount Sinai, Becker’s Hospital Review, Healthcare IT News and HIT Consultant, marks OpenEvidence’s first enterprise-scale deployment extended to the full clinical care team rather than physicians alone.
Who Gets Access, and How It Works
The rollout covers physicians, registered nurses and pharmacists throughout Mount Sinai’s network — a notably broader rollout than typical physician-only clinical decision support tools. Inside the EHR interface, a clinician can type a natural-language question about a drug interaction, a diagnostic question, or a treatment guideline and receive a synthesized, citation-backed answer grounded in established clinical guidelines and peer-reviewed literature, according to Mount Sinai’s own announcement and coverage from HLTH and OpenEvidence itself. The platform is designed to function less like a general-purpose chatbot and more like a governed medical search engine, restricting its outputs to sourced, verifiable material specifically to reduce the risk of the kind of unsupported or fabricated answers that have dogged consumer-facing generative AI tools.
Why the EHR Integration Matters More Than the AI Itself
OpenEvidence and similar evidence-lookup tools have existed as standalone products for a while; what’s new here is the direct embedding into Epic, the EHR system that already anchors most clinicians’ daily workflow. Health-IT reporting on the deal frames this as addressing one of the most persistent barriers to AI adoption in hospitals: clinicians who are already stretched thin have historically been reluctant to open a separate application or browser tab to consult a decision-support tool mid-shift. By putting the natural-language query function inside the same screen where a doctor or nurse is already documenting care, Mount Sinai and OpenEvidence are betting that usage rates will be far higher than they would be for a bolt-on tool.
Part of a Broader 2026 Enterprise AI Wave
Mount Sinai’s deal is one of at least 15 enterprise-level AI agreements signed by U.S. health systems in 2026 tracked by Becker’s Hospital Review, alongside deployments such as University of Rochester Medicine’s enterprise AI platform agreement with Qualified Health across its eight-hospital, roughly $6 billion revenue system. Industry trackers cited in that reporting put the number of FDA-cleared or -authorized AI and machine-learning-enabled medical devices above 1,500 by 2026, up from roughly 1,250 in mid-2025 — underscoring how quickly AI has moved from pilot projects into infrastructure treated as a standard part of hospital operations rather than an experimental add-on.
The Skeptical Read
Even a citation-grounded tool like OpenEvidence raises questions that health-IT analysts and clinicians have pressed on repeatedly through 2026: how often do the underlying citations actually support the specific clinical claim being made, who audits that alignment over time as guidelines change, and does easy in-workflow access to an AI answer risk narrowing or homogenizing clinical judgment rather than simply speeding up literature review clinicians would have done anyway. Broader surveys of hospital AI adoption this year have also found a persistent gap between enterprise licensing announcements and steady frontline usage — a tool being switched on for every clinician at a hospital is not the same as every clinician actually relying on it daily, and Mount Sinai’s own results on sustained usage rates have not yet been publicly reported.
What’s Next
The real test for Mount Sinai’s rollout will be usage data over the coming months: how many of the health system’s physicians, nurses and pharmacists actually query the tool regularly, and whether hospital leadership or independent researchers publish any data on whether it changes clinical decisions, reduces errors, or simply saves clinicians time looking things up they would have found anyway through traditional reference tools like UpToDate. If the deployment is judged a success, it is likely to accelerate similar EHR-embedded AI deals at other academic medical centers watching closely — several of which, like Rochester’s Qualified Health agreement, are already underway. The broader question hovering over all of 2026’s enterprise AI deals, healthcare-IT analysts note, is whether hospitals are building genuine clinical value into their workflows or simply keeping pace with a competitive market in which not having an AI strategy has become its own kind of institutional risk.
