Tairex unveils multi-agent AI system to work alongside physicians
Tairex, an AI healthcare technology company incubated by the Institute for AI Industry Research (AIR) at Tsinghua University, has unveiled a system in which specialized AI agents work alongside physicians on clinical cases. The Multidisciplinary Human-Agent Collaboration (MHAC) Ecosystem for Agent Hospital was presented on 21 October.
- Tairex, incubated by Tsinghua's AIR, unveiled the MHAC Ecosystem for Agent Hospital on 21 October.
- The system has 54 AI doctors covering more than 20 clinical specialties, trained against millions of AI patients.
- Tairex says the AI doctors reached over 96% accuracy on the MedQA dataset.
- Agents work simultaneously on multi-specialty cases and highlight disagreements for the physician.
- The company says AI supports decisions rather than making critical ones independently.
Trained in a simulated hospital
According to Tairex, the system is built through training in a simulated environment populated by millions of highly realistic AI patients. In that setting, an AI doctor can conduct tens of thousands of simulated consultations daily.
The company says this training yields a system of 54 AI doctors and millions of AI patients, covering more than 20 clinical specialties and tens of thousands of diseases. Tairex states that the AI doctors have achieved over 96% accuracy on the MedQA dataset.
Agents that collaborate and flag disagreements
When a patient's condition spans several specialties, the system dynamically activates the relevant agents to work at the same time. Instead of operating in silos, the agents present their clinical reasoning and evidence on a unified interface. Disagreements between agents are highlighted for the human physician to review.
Tairex gives an example from the system. An endocrinology agent proposed a beta-blocker for a patient with palpitations and a history of asthma. A respiratory medicine agent then flagged that non-selective beta-blockers could trigger bronchoconstriction and suggested considering a selective beta-1 blocker.
Because several agents can give recommendations at once within the clinical interface, doctors can consider cross-specialty perspectives without waiting for a separate meeting, according to the company.
Access, data security and the role of the physician
Tairex says AI agents modeled on specialists could support multiple departments and potentially reach primary care facilities, which would help make high-quality medical resources more widely accessible. The company stresses that AI in this model is designed to improve efficiency and support clinical decision-making rather than independently make critical decisions.
On data protection, Tairex says its underlying products take reference from the EU's General Data Protection Regulation (GDPR) and South Korea's Personal Information Protection Act (PIPA), with controls governing the handling and transfer of sensitive data fields. The company has also obtained China's Multi-Level Protection Scheme (MLPS) Level 3 certification.
What it means for healthcare institutions
Tairex says the ecosystem aims to help high-quality medical resources reach more healthcare institutions, including those at the primary care level. The company frames this as bringing more equitable access to healthcare closer to reality.
The announcement does not give details on pricing, deployment timelines or clinical trials, and the accuracy figure comes from the company itself. For hospitals, the central design choice is that physicians stay in charge: the agents surface reasoning and conflicts, and the human makes the call.
