Stanford's Biomni AI agent autonomously handles complex biomedical research across 10,000+ labs

Biomni AI Agent Redefines Autonomous Biomedical Research

Stanford University researchers have introduced Biomni, an AI agent designed to autonomously execute diverse biomedical research tasks, marking a significant shift in how scientists approach discovery2

2

. Published in the journal Science, this AI-powered biomedical co-scientist goes far beyond conventional chatbots by integrating large language model reasoning with retrieval-augmented planning and code-based execution to automate complex biomedical research workflows1

1

. The system can read literature, form hypotheses, select datasets and tools, write code, interpret results, and suggest next-stage experiments in complete research workflows. A prototype is already in use by more than 10,000 labs across academia and industry, making it the most widely used AI co-scientist system in biomedicine2

2

.

Building a Unified Biomedical Action Space

What distinguishes Biomni from other AI systems is its comprehensive mapping of the biomedical action space. The platform's action-discovery agent mines tools, databases, and protocols from thousands of publications across 25 domains defined by bioRxiv, creating a unified agentic environment1

1

. Biomni incorporates 150 specialized biomedical tools, 105 software packages, and 59 databases spanning all 25 biomedical subdomains, ranging from genetics to neurology2

2

. "If you think of an agent as a carpenter, a carpenter without tools is just a carpenter who can talk," explained Jure Leskovec, the Alfred and Rebecca Lin Professor of computer science at Stanford and senior author of the paper. "With Biomni, we give the carpenter a set of tools so it can build."2

2

Accelerate Scientific Discovery Through Task Generalization

Systematic benchmarking demonstrates Biomni's strong generalization across heterogeneous tasks including causal gene prioritization, drug repurposing, rare-disease diagnosis, microbiome analysis, and molecular cloning, all without task-specific tuning1

1

. In one real-world case study, a user uploaded more than 450 files of continuous glucose monitoring, food intake, and physical activity data, asking Biomni to analyze the data and find interesting hypotheses. In just 40 minutes, the system cleaned and unified the data, generated visualizations, and identified patterns relating food intake and body temperature—work that Leskovec estimates would have taken 60 or more hours for a human to complete2

2

. Real-world case studies also show Biomni interpreting multi-modal datasets, optimizing protein stability, orchestrating wet-lab instruments, and generating experimentally testable protocols1

1

.

AI Augmenting Human Scientists, Not Replacing Them

The researchers emphasize that Biomni addresses a critical bottleneck in biomedical research: the inverse relationship between scientific information and the pace of discovery. As the volume of knowledge, data, and tools has grown, innovation has slowed because developing even a single hypothesis requires substantial investment in reading literature, ingesting datasets, writing code, and searching for unexplored patterns2

2

. "The hurdle in biomedical science is not intelligence or ideas; it is mechanics," Leskovec noted. "It's this laborious stuff that slows innovation. Biomni can do this work in minutes."2

2

. Importantly, Biomni provides full citations and tracking of its work, enhancing reproducibility and making science more rigorous2

2

. Kexin Huang, a former doctoral student in Leskovec's lab who led the development, emphasized that "this is not about machines taking over science, but more about machines becoming a powerful new partner to augment human researchers."2

2