OpenEvidence Workflow Ownership in Medicine
OpenEvidence
The real moat in medical AI is not the model, it is the closed loop between trusted content, daily clinician usage, and workflow data. OpenEvidence got early distribution by being free and avoiding patient records, then turned that usage into exclusive journal partnerships and now product expansion into scribes and calling. That makes it harder for a general model company to copy, because the advantage sits in where doctors work and what evidence they trust, not just in raw reasoning quality.
-
OpenEvidence started with a simple wedge, clinical search for doctors, but the bigger asset is the traffic. By mid 2025 it had exclusive partnerships with NEJM and JAMA, and by late 2025 it was monetizing roughly 45% of U.S. physicians through high value pharma and device ads shown at the exact moment a doctor is researching a case.
-
This is different from Hippocratic AI, which sells agent hours to health systems for non diagnostic patient interactions, and different from Harvey and Hebbia, which sell expensive workflow software plus implementation help into law and finance. The shared pattern is that each is wrapping general models with proprietary data, domain workflows, and high touch deployment.
-
The next step is where the moat compounds. Once OpenEvidence moves from search into note drafting, EHR connectivity, drug checks, and outbound calling, every interaction creates more workflow level data about what doctors ask, what outputs they accept, and where AI can save time. That pushes the product from reference tool toward system of action.
The category is heading toward workflow ownership inside each profession. In medicine, the winners are likely to be the products that start with trusted answers, then expand into the daily tasks around diagnosis and treatment. If OpenEvidence keeps turning clinician attention into proprietary data and deeper workflow integration, it can grow from a free reference app into a core interface for doctor productivity.