MGI Tech and Shanghai AI Laboratory Unveil ProtoPilot and BioLab Bench, Pioneering Physical AI for Life Sciences

In a significant advancement for the intersection of artificial intelligence and life sciences, MGI's subsidiary Genoria AI, in collaboration with the Shanghai Artificial Intelligence Laboratory, has unveiled two groundbreaking innovations: ProtoPilot and BioLab Bench. These tools aim to bridge the gap between digital intelligence and physical execution in biological research, marking a pivotal moment in the development of Physical AI. The announcement was made on July 5, 2026, and the research underpinning these innovations was published as a preprint on arXiv in June 2026.

Innovations in Physical AI

ProtoPilot is a self-evolving multi-agent system designed to manage the entire experimental lifecycle. It operates through a series of stages: Design2Protocol, Protocol2Code, Device Execution, and Wet-Lab Feedback. This sophisticated system is not just about generating theoretical answers; it actively learns from failures encountered during experiments. For instance, when a PCA assembly step failed due to antibiotic resistance screening issues, ProtoPilot was able to diagnose the problem and autonomously regenerate a corrected protocol. This capability underscores the emergence of true Physical AI, which can adapt and improve based on real-world laboratory scenarios.

In conjunction with ProtoPilot, BioLab Bench serves as the first comprehensive evaluation framework in the industry. Unlike traditional evaluation systems that focus solely on whether an AI agent provides correct answers, BioLab Bench assesses whether these agents can execute tasks on actual automation equipment. This new standard is crucial for ensuring that AI can translate theoretical knowledge into practical applications, a necessity in the fast-evolving field of life sciences.

The implications of these innovations extend beyond mere academic interest. With the introduction of BioAgents, which will evolve not just through text-based training but through a continuous Physical AI experimental loop, the landscape of laboratory research is set to change dramatically. These agents will accumulate data from real research tasks, automation operations, expert validations, and failure cases, enabling them to develop integrated reasoning, execution, and validation capabilities. The ultimate goal is to facilitate unattended intelligent laboratories that can operate 24/7, significantly enhancing productivity and efficiency in scientific research.

The Path to AI for Science

MGI's exploration into AI began in 2019, culminating in the establishment of Genoria AI in April 2026, which focuses specifically on AI for Science (AI4S). This initiative is geared towards creating a closed-loop infrastructure for life sciences that integrates dry and wet lab processes. The groundwork for this was laid in 2025 when Dr. Yang Meng, the Chief AI Officer of MGI, collaborated with Professor Nattiya Hirankarn from Chulalongkorn University to publish a paper in Nature Biomedical Engineering. This paper introduced "PrimeGen," a collaborative multi-agent system that streamlined primer design, experimental validation, and automated workstation execution.

Dr. Yang Meng, now serving as CEO of Genoria AI, articulated a vision that diverges from the typical race for computational power seen in many leading AI companies. Instead of relying solely on scaling compute resources, MGI emphasizes agent scaling and closed-loop data engineering. This approach organizes real-world tasks, device constraints, expert feedback, and laboratory results into a dynamic training environment where AI can continuously evolve and adapt.

The collaboration with the Shanghai AI Laboratory, which was officially unveiled at the World AI Conference in July 2020, further strengthens this initiative. The laboratory aims to position itself as a leading research institute, focusing on original theories and key technologies in AI. By attracting top talent and fostering a collaborative research environment, it seeks to make substantial contributions to the fields of industry, healthcare, and education.

Why it matters

The introduction of ProtoPilot and BioLab Bench represents a significant leap forward in the application of AI within the life sciences. For creators and technologists, this development highlights the potential for AI to not only assist in research but to fundamentally transform how experiments are designed, executed, and validated. The ability of AI agents to learn from real-world failures and successes will lead to more robust and reliable experimental outcomes.

Moreover, as BioAgents become capable of operating autonomously in laboratories, the implications for efficiency and productivity are profound. Researchers will be able to focus on higher-level problem-solving and innovation, while the AI handles routine tasks and optimizes experimental protocols. This shift could accelerate the pace of discovery in life sciences, making it imperative for creators and technologists to adapt and integrate these advancements into their workflows.

In conclusion, the launch of ProtoPilot and BioLab Bench not only sets a new standard for AI in life sciences but also opens the door for a future where intelligent systems play a central role in scientific research. As these technologies evolve, they promise to redefine the boundaries of what is possible in the realm of experimental biology.