Recently, Tairex, incubated by Tsinghua University's Institute for AI Industry Research (AIR), officially launched its multi-agent collaborative ecosystem.
According to the introduction, this ecosystem breaks through the limitations of traditional single AI tools by building multiple doctor avatar agents with different specialty backgrounds, enabling AI to achieve real-time cross-specialty collaboration and multidisciplinary consultations during diagnosis and treatment.
Notably, the internationally authoritative academic journal Nature previously published an article reporting on the Agent Hospital (Tairex AI Hospital) built by the Tsinghua University team as a core case of Chinese AI Agent implementation in the medical field.
Tairex Chief Marketing Officer Li Zheng stated that what Tairex has built is a complete diagnosis and treatment system, as well as three-dimensional regional hierarchical diagnosis and treatment capabilities.
He hopes that through the grassroots invocation of clinical expert avatars, grassroots village doctors or community doctors can form a three-dimensional service system with higher-level doctors, achieving the goal of "minor illnesses treated within the county, major illnesses treated within the province."
Tairex Chief Product Officer Qiao Yuchen further pointed out that to achieve this goal, AI must not only understand medicine, but also learn to collaborate like human doctors.
"Our multi-avatar agents will not only undergo massive training in the virtual world, but will also be deeply bound with doctors in real diagnosis and treatment, evolving alongside them. When facing complex cross-department conditions, doctor avatars from different specialties can 'team up' at any time to provide decision-making references, while the final judgment and responsibility always remain in the hands of human doctors."
Currently, Tairex has built 54 AI doctors and millions of ultra-realistic AI patients, covering more than 20 clinical departments and tens of thousands of diseases. The AI doctors' test accuracy on the MedQA dataset exceeds 96%.
According to data released by Tairex, on average, after each doctor avatar agent collaborates with a doctor on approximately 200 cases, the adoption rate of its diagnosis and treatment recommendations can steadily reach over 90%.
Qiao Yuchen introduced that Tairex's multi-agent collaborative ecosystem can arrange high-level expert avatars from the same department to provide guidance, or directly organize multidisciplinary consultations.
In this ecosystem, there is no limit to the number of expert avatars, suggestions from different avatars are presented simultaneously, inconsistencies are highlighted with reminders, and the final comprehensive judgment is made by the doctor.
This mechanism breaks the limitations of physical space, allowing high-quality medical experience to cross departments and benefit patients.
At the same time, grassroots doctors themselves can also have their own dedicated avatars.
Li Zheng stated that with the help of Tairex's ecosystem, a grassroots doctor is expected to manage thousands of residents within their jurisdiction more efficiently. Their avatar can monitor data fluctuations transmitted back by wearable devices in real time, determine which indicators need early warning, which populations need referral to higher-level hospitals, and provide residents with personalized health management plans.
As of July 2026, Beijing Tsinghua Changgung Hospital has had 18 clinical department outpatient clinics test-run the Agent Hospital system, with more than 40 doctors using it regularly. According to outpatient process time statistics, doctors' overall diagnosis and treatment efficiency has improved by 30%.
Comments