In February 2025, OpenEvidence announced a new $75 million injection of cash from a Series A funding round with Sequoia Capital, helping the startup reach unicorn status – a $1 billion valuation. The startup, founded by a Harvard PhD and Kensho founder, works to integrate artificial intelligence into the world of medicine. The company’s product works as a diagnostic tool, pulling information from peer reviewed journals to assist in the investigative process of medicine using the best available scientific evidence. Blazoned on the company’s website, OpenEvidence boasts widespread use–advertising that over 10,000 care centers actively utilize the technology. Backed by some of the biggest names in the game, OpenEvidence could be the difference maker in diagnostic medicine.
History of OpenEvidence
Led by two technological visionaries, OpenEvidence has assembled a team of highly qualified engineers to create what appears to be the best AI diagnostic tool on the market. Daniel Nadler, the CEO and Founder, earned both a bachelor's and a PhD from Harvard University in math-related fields. His first venture, Kensho, was a quantitative trading tool that was sold to S&P Global (NYSE: SPGI) for $550 million in 2018. Co-founder and CTO has similar credentials, attending both Cornell and Harvard. OpenEvidence’s website corroborates the integration of medicine and artificial intelligence, showing the strong team of technical experts and medical leaders who work to make the product effective.
Having emerged from the Mayo Clinic’s Platform Accelerate Program, OpenEvidence utilizes peer-reviewed sources like JAMA Network and NEJM. Having received early backing from top VC firms, like Sequoia Capital, Google Ventures, and Kleiner Perkins, OpenEvidence has flourished with fundraising reaching more than $100 million. Furthermore, the company has made strategic partnerships with top names in the medical research industry, making multi-year agreements with JAMA, NEJM, and Elsevier’s ClinicalKey AI.
OpenEvidence’s allure is found not only in its medical application but also in its user-friendly interface. The application, which has been described as a “medical copilot”, now boasts a user population above 100 thousand physicians, which accounts for about 25 percent of the US doctor population. Lauded for many reasons, physicians report that OpenEvidence’s product bypasses hospital bureaucracy and aids them rather than replaces them.
Technical Background
OpenEvidence’s system is built upon vertical Large Language Models (LLMs), which are AI models that are specifically trained for medical decision making. The model focuses on perfecting medical explanations and accuracy, pulling from highly touted, peer-reviewed journals like the New England Journal of Medicine and JAMA. The core innovation in OpenEvidence’s technology, though, is its ability to decline to answer when there is no research supporting a specific claim. This innovative feature ensures that the AI avoids a phenomenon called “hallucination”–when AI outputs non-factual information when there is no available information to support a claim. Thus, OpenEvidence provides a citation for every claim, acting as a customer-facing quality assurance mechanism. Furthermore, the startup alleviates privacy concerns, assuring that all patient information is discarded.
Overview: AI’s Role in Healthcare
Artificial Intelligence is no longer a futuristic promise in health care–it is actively being applied on a daily and widespread basis. AI has been making strides in pattern recognition based medicine–like detecting tumors in radiology and reading ECGs in cardiology. Surprisingly, in such fields, AI has not been far behind physicians, with an NIH study finding that AI was about 86% accurate in detecting heart arrhythmias through the reading of ECGs. Due to AI’s ability to compile massive amounts of patient data, research, and experimental data, it makes a strong tool for physicians as a clinical decision support system (CDSS). For example, in an article published by Dartmouth University about a recent symposium at the institution, zebraMD is showcased, explaining that the application can utilize thousands of medical files to provide insights into rare disease diagnosis.
AI’s utilization spans far beyond diagnosis, however. AI scribes automate patient documentation, enabling efficient data collection, reduced manual work, and increased time spent on treatment. Furthermore, AI’s computational power is reported to be instrumental in predicting risk for specific conditions and adopting treatments in a more timely fashion.
Ethical Concerns
Although AI seems to be ushering in a new and efficient age for medical diagnostics, there are clear ethical roadblocks that must be alleviated before a full implementation is possible. First and foremost, medical data privacy is of the utmost importance, as outlined by guidelines like HIPAA. When working with technology, the risk of data breach is always a possibility—thus, for a product like OpenEvidence to be implemented on a wide scale, cybersecurity procedures and technology would need to be nearly impermeable.
Aside from privacy concerns, many have expressed qualms regarding unintended bias in AI medical systems. Some contend that underrepresented populations may be misdiagnosed due to a lack of patient data. Furthermore, many have expressed concerns that biases within these systems can lead to misdiagnosis in mental health-related cases, which can certainly lead to patient self-harm.
Additionally, AI models do not always give exact reasoning for their decisions and recommendations. While that may not be the case with OpenEvidence, it is prevalent in other Large Language Models.
Despite what many fear-mongering technology influencers may say on social media, AI is not replacing clinicians; rather, it is making physicians more efficient and allowing them to create treatment plans promptly using peer-reviewed research. Furthermore, the very human aspect of clinical treatment will never be replaced, as patients will always need the warmth provided by a well-versed doctor—diligence on clinical decisions will never be fully allotted to a machine. Nevertheless, AI has been a market disruptor across industries and geopolitical borders, now making its way into one of the most human-centered fields: healthcare.
Whether OpenEvidence becomes a hallmark of the diagnostic process or it's just the spark which ushers in the AI evolution in medicine, one conclusion is evident: AI has scrubbed into the OR and it is not leaving anytime soon.