OpenEvidence and the Endodontic Classroom

By Tung Bui, DDS, FICD

Artificial intelligence is no longer an abstract concept discussed in computer science seminars; it is a tool that is already changing how we teach and practice endodontics. When I mentor AEGD residents about evidence based practice, I often begin by showing how many papers are published each year and how little time we have between patients to review them. This flood of information is not a problem that better memory can solve. It requires technology that helps us find trustworthy evidence and presents it in a way that clinicians can use during a busy clinic session. OpenEvidence is one of the most interesting responses to that challenge. It is a platform built for clinicians, and it rewards the curiosity of educators and students who want to ground their teaching in current research.

What is OpenEvidence?

OpenEvidence is an artificial intelligence powered medical information platform that draws exclusively from peer reviewed medical literature. Instead of scraping the entire internet, the software indexes metadata from PubMed abstracts, full text journal articles, monographs and book chapters. When a user, who must be a verified healthcare professional, asks a clinical question, OpenEvidence identifies potentially relevant sources, selects the most authoritative papers based on relevance, publication date, journal impact factor and citation count, and then synthesizes a short answer with clickable citations. This output arrives within seconds, allowing a clinician to remain in the loop and dig deeper if needed. To limit hallucinations, the system abstains when evidence is inconclusive and always includes references.

The platform is intentionally restricted to healthcare professionals. Physicians, dentists, nurses and other clinicians verify their status by entering a National Provider Identifier or similar credentials. Medical and dental students can register by submitting proof of enrollment. Non-clinicians are permitted only two searches per day, and there is currently no way for the public to purchase unrestricted access. This gatekeeping maintains quality by ensuring that the user community understands the consequences of acting on clinical information. OpenEvidence complies with the Health Insurance Portability and Accountability Act (HIPAA) and will sign a Business Associate Agreement with covered entities, but it warns users not to enter protected health information.

Origin Story and Business Model

OpenEvidence was co founded by Daniel Nadler and Zachary Ziegler. Nadler is a Harvard trained economist and artificial intelligence entrepreneur who previously founded Kensho, a financial AI firm. During the COVID‑19 pandemic he observed that physicians were drowning in a fire hose of literature where medical knowledge doubles roughly every seventy three days. He assembled a team of researchers from Harvard and MIT, including Ziegler, to build smaller, highly specialized models trained solely on peer reviewed medical literature. The company launched in 2021 under the Mayo Clinic Platform Accelerate program and released its direct to clinician application in 2022.

The go to market strategy treats doctors like consumers rather than hospital administrators. OpenEvidence is free for United States clinicians and students and generates revenue through targeted advertising. The company has raised several rounds of funding: self financed in 2021, a friends and family round in 2023, a seventy five million dollar Series A led by Sequoia Capital in February 2025, and a two hundred and ten million dollar Series B co led by Google Ventures and Kleiner Perkins in July 2025, valuing the firm at 3.5 billion dollars. Those private investments mean there is no publicly traded stock; interested investors would need access to future private rounds, which are typically limited to accredited investors. The rapid adoption with over forty percent of U.S. physicians logging in daily, is proof that the free, advertising supported model can scale. Disclosure; I have been working patiently with my hedge fund manager to acquire private shares of OpenEvidence.

How does it work?  

OpenEvidence’s core function is known informally as Ask. Clinicians type or speak a question into a mobile or desktop interface, such as “write home care instructions for apical surgery” or “what is the dosing of Penicillin VK for a ten year old. The platform uses natural language processing to interpret the question and then searches its indexed corpus of more than thirty five million peer reviewed publications. It ranks the articles, extracts the relevant facts and composes a short answer with inline citations. The interface invites follow up questions and may suggest related queries to help refine the search. When viewing citations, clinicians can expand a details button to read a summary of each reference, rate the helpfulness of the response, copy a shareable link and see automatically generated follow on questions. The platform warns users to verify that citations actually answer the question, reinforcing healthy skepticism.

While the standard search is designed for rapid answers, some questions require deeper research. Deep Consult is an AI agent that autonomously reads hundreds of studies and produces a longer research brief. The service runs complex computations, more than a hundred times the compute of a standard search, but remains free for verified clinicians. Deep Consult is useful when preparing literature reviews or when dealing with complex medical histories. For example, when a patient presents to the clinic with a complex medical history, we can request a Deep Consult on the list of conditions and receive a summary of all available evidence, including randomized controlled trials and systematic reviews.

Visits is a digital clinical assistant introduced in 2025. It acts like a medical scribe: recording patient encounters, drafting notes and enriching the assessment and plan with guidelines and current research. Visits allows clinicians to ask questions using the patient’s full history and documentation, organizes patient files into a searchable repository and generates polished notes that can be pasted into the electronic health record. For dental educators supervising residents, this module can lighten the administrative load and model best practice documentation.

The Dialer module provides a HIPAA secure phone line with unlimited minutes and smart caller ID. When calling a patient after hours, the clinician’s personal number is masked, and the call can be recorded or linked to a Visit. The application also offers clinical trial matching so providers can identify active trials for conditions like cemental tears, external invasive cervical resorption, or regenerative endodontics procedures. Inbox, a secure messaging centre, allows clinicians to manage queries and results from the platform, while Discover curates featured stories and recent advances from journals such as The New England Journal of Medicine and JAMA. Looking into the future, one could customize the feed to focus on relevant endodontic topics. These feeds transform idle scrolling into a learning opportunity.

Accessing the service

Downloading OpenEvidence is straightforward: search for the name in the Apple App Store or Google Play, or use the web version (https://www.openevidence.com). After installation, users must create an account. Physicians, dentists and allied providers verify their identity by entering an NPI; dental students and endodontic residents upload proof of enrollment. The platform is free for verified professionals, but non clinicians are limited to two searches per day. Institutions seeking to integrate Visits or store protected health information must sign a Business Associate Agreement [4]. The platform’s terms permit the company to collect usage data and sell anonymized, non personal information for commercial purposes, so educators should remind residents not to enter identifiable patient data.

How OpenEvidence benefits dental educators and residents

The everyday functions of Ask, Deep Consult and Visits make OpenEvidence a natural companion for endodontic education. Residents often struggle to connect pathophysiology with evidence based management. With Ask, a resident confronted with burning mouth syndrome in an endodontic residency program can pose a question like “what are the etiologic factors and management options for burning mouth syndrome” and receive a concise summary with citations that they can read before presenting the case to faculty. When our residents debated whether every patient needs a pre-operative cone beam computed tomography scan, Deep Consult produced a structured report summarizing guidelines, systematic reviews and cost effectiveness analyses. That report grounded our discussion in evidence rather than anecdotes and allowed us to model critical appraisal skills.

OpenEvidence also shines when we evaluate new technology. Sales representatives often pitch devices like GentleWave with claims of superior cleaning and less postoperative pain. By asking the platform for a Deep Consult on the device, we quickly learn whether independent trials support those claims. If the evidence is weak or only animal studies exist, we can protect our budgets and our patients. Similarly, when a patient asks whether a muscle relaxant we plan to prescribe interacts with their antiretroviral medications, the platform surfaces drug monographs and interaction studies. Instead of relying on memory or generic interaction checkers, we see the primary literature.

Educators can integrate OpenEvidence into lesson plans. Assign students to craft endodontic questions and evaluate the AI’s responses. Users can filter searches by All, Guidelines and Standard of Care or Clinical Evidence, with results that are tagged as Highly Relevant, Leading Journal or New Research. Those tags help students judge the strength of evidence and understand that not all papers are equal. Encourage residents to follow the citations, read the original articles and consider the applicability to endodontic practice.  Remind them that OpenEvidence is an experimental tool and does not replace critical thinking.

Why use a purpose built platform instead of general large language models?

General purpose large language models such as ChatGPT are trained on vast swaths of the internet, including blogs, social media and satirical sites.  Daniel Nadler argues that “an index of websites is not an index of facts”. Because mainstream models do not differentiate between high quality medical research and unreliable sources, they are prone to error and hallucinations. In contrast, OpenEvidence is trained on peer reviewed publications and selects citations based on relevance and impact. The platform abstains when evidence is insufficient, and it displays citations so clinicians can verify the information. The general public without an NPI is limited to two searches per day; this gatekeeping reduces misuse and encourages proper clinical oversight. In other words, the platform is designed not just to answer questions but to foster evidence based reasoning.

Looking ahead

OpenEvidence’s vision extends beyond answering questions. The company plans to expand globally and develop advanced AI models capable of sophisticated diagnostic reasoning and personalized treatment recommendations. Future modules may integrate with electronic health records to surface evidence during charting and provide application programming interfaces for third party applications. The platform already offers clinical trial matching and is experimenting with ways to help clinicians draft prior authorization letters and patient education handouts. These innovations suggest that artificial intelligence will become an ambient part of clinical workflows, not a separate destination.

A measured embrace of AI in endodontics

For endodontic educators, AI is not a replacement for mentorship but a catalyst for deeper learning. OpenEvidence demonstrates how a tool built on high quality data, designed for clinicians, can shrink the distance between the literature and the operatory. Whether we are confronting an unusual neuropathic pain condition, debating imaging protocols or evaluating a flashy new device, we can turn to a platform that delivers rapid, referenced answers and encourages us to read the source material. By integrating these tools into our teaching, we prepare residents for a future in which evidence is at their fingertips and critical appraisal is more important than memorization. Artificial intelligence in education is not about abdicating judgement; it is about augmenting our ability to find and apply the best available evidence.

References

1. FeldmanA, Shrivastava R. This AI founder became a billionaire by building ChatGPT for doctors. Forbes. July 15, 2025. https://www.forbes.com/sites/amyfeldman/2025/07/15/this-ai-founder-became-a-billionaire-by-building-chatgpt-for-doctors/
2. OpenEvidence, the fastest‑growing application for physicians in history, announces $210 million round at $3.5 billion valuation.PR Newswire. July 15, 2025. https://www.prnewswire.com/news-releases/openevidence-the-fastest-growing-application-for-physicians-in-history-announces-210-million-round-at-3-5-billion-valuation-302505806.html
3. Jiang Y. OpenEvidence and the future of AI medical assistants. VerticalAI Newsletter. 2025. https://www.newsletter.lukesophinos.com/p/135-openevidence-vertical-ai-for