The Stanford-Princeton AI Coscientist Team has launched MedOS, the first AI-XR-cobot system designed to actively assist clinicians in real clinical environments. This innovative platform combines smart glasses, robotic arms, and multi-agent artificial intelligence to function as a real-time co-pilot for doctors and nurses, aiming to reduce medical errors, accelerate precision care, and support overburdened clinical teams.
Physician burnout has reached crisis levels, with over 60% of U.S. doctors reporting symptoms according to recent studies. MedOS addresses this challenge not by replacing clinicians but by reducing cognitive overload, catching errors, and extending precision through intelligent automation and robotic assistance. The system builds on years of innovation from the team's previous breakthrough, the LabOS, bridging digital diagnostics with physical action across operating rooms and bedside diagnostics.
MedOS introduces a "World Model for Medicine" that combines perception, intervention, and simulation into a continuous feedback loop. Using smart glasses and robotic arms, it can understand complex clinical scenes, plan procedures, and execute them in close collaboration with clinicians. The platform has demonstrated early promise in tasks including laparoscopic assistance, anatomical mapping, and treatment planning. In surgical simulations, MedOS has shown the ability to interpret real-time video from smart glasses, identify anatomical structures, and assist with robotic tool alignment, functioning as a true clinical co-pilot.
The system's breakthrough capabilities include a multi-agent AI architecture that mirrors clinical reasoning logic, synthesizes evidence, and manages procedures in real time. MedOS achieved 97% accuracy on MedQA (USMLE) and 94% on GPQA, outperforming frontier AI models like Gemini-3 Pro, GPT-5.2 Thinking, and Claude 4.5 Opus. The platform also utilizes MedSuperVision, the largest open-source medical video dataset featuring more than 85,000 minutes of surgical footage from 1,882 clinical experts available at https://medos-ai.github.io/.
Clinical testing has demonstrated significant improvements in performance metrics. Registered nurses improved from 49% to 77% accuracy with MedOS assistance, while medical students improved from 72% to 91% accuracy in fatigue-prone environments. The system has also shown capability in identifying important clinical patterns, including uncovering immune side effects of the GLP-1 agonist Semaglutide (Wegovy) from FDA databases and identifying prognostic implications of driver gene co-mutations on cancer patients' survival.
Dr. Le Cong, leader of the Stanford-Princeton AI Coscientist Team and Associate Professor at Stanford University, emphasized that "the goal is not to replace doctors. It is to amplify their intelligence, extend their abilities, and reduce the risks posed by fatigue, oversight, or complexity." Dr. Mengdi Wang, co-leader of the collaboration, noted that "MedOS reflects a convergence of multi-agent reasoning, human-centered robotics, and XR interfaces" designed to help clinicians manage complexity in real time.
MedOS is launching with support from NVIDIA, AI4Science, and Nebius, and has been deployed in early pilots. The system will be showcased at a Stanford event in early March, followed by a public unveiling at the NVIDIA GTC conference in March 2026, with session information available at https://www.nvidia.com/gtc/session-catalog/sessions/gtc26-s81748/. Additional information about the project can be found at the official site https://ai4medos.com/.
