About 650,000 American physicians — roughly two-thirds of the country's doctors — now use a chatbot called OpenEvidence to help decide how to treat their patients, the company told NBC News. In April alone, the service was queried during nearly 27 million clinical encounters, plus another 1.2 million users abroad. A year ago, the startup was worth $1 billion; today it is worth $12 billion, with Sequoia, Google Ventures, Nvidia, and Andreessen Horowitz on the cap table.
Sign-up requires the unique NPI number that the federal government issues to licensed providers, and the core product is free, funded by ads from drug and medical-device companies. To generate its answers, OpenEvidence licenses content from top journals like NEJM and JAMA. About 60% of searches are clinical decisions: a physician describes a patient's profile and other conditions and asks what to do, according to Dr. Anupam Jena, a Mass General internist who is currently sifting through 90 million queries that doctors have sent the tool since 2024.
Jared Perlo reports for NBC News that an unpublished December study found OpenEvidence got the right answer fewer than 45 times out of 100 on tougher medical questions, despite scoring a perfect 100% on the U.S. Medical Licensing Examination. MaineHealth tells its doctors not to enter protected health information into the chatbot. A midcareer doctor in Missouri said students who lean on the tool from day one are already losing their ability to tell signal from noise.
Meanwhile, in March, Mount Sinai wired OpenEvidence directly into the electronic health record portal used by its 47,000 employees.
Previously:
