Stanford Scientists Construct an AI Lab Associate

For many scientists who research biology and medication, breakthroughs rely on advanced lab experiments, large datasets, and a rising variety of specialised software program instruments. This work usually entails repetitive duties and fragmented procedures that gradual progress. 

A crew of Stanford students noticed a solution to speed up biomedical discovery by making a digital AI biologist that operates alongside human scientists. Funded partly by a Stanford HAI Hoffman-Yee Research Grant, the crew constructed Biomni, a general-purpose biomedical AI agent that may execute a variety of biomedical analysis duties. 

The open-source mission incorporates experience from the Stanford departments of Computer Science, Genetics, Pathology, Medicine, and Pediatrics – in addition to contributions from Genentech, Arc Institute, the University of Washington, and the University of California, San Francisco.

“Today, we have abundant biomedical data, but we don’t have enough human researchers to analyze it all,” says Kexin Huang, a former Stanford PhD candidate and one of many co-creators of Biomni. “While human biologists are limited by specialized expertise, AI can integrate across disciplines and manage thousands of concurrent tasks.”

A Superpower for Human Scientists

Biomni is a cloud-based platform that consists of two fundamental elements: a digital work surroundings, the place all actions associated to a analysis activity are dealt with, and an agentic structure that allows the AI biologist to carry out new duties that it hasn’t encountered earlier than with out further coaching. The platform combines massive language fashions with greater than 150 specialised bioinformatics instruments; 59 curated databases containing protein constructions, genomic variants, and literature repositories; and greater than 100 software program packages for molecular modeling, single-cell evaluation, and the like.

To collaborate with Biomni, the scientist enters a question into the agent’s chat window. For instance, the consumer may start with an open-ended request, comparable to: “Analyze the attached Perturb-seq data and generate a meaningful hypothesis.” The agent begins by discovering essentially the most related instruments wanted to finish the duty. Next, it applies LLM-based reasoning and its understanding of the biomedical subject to formulate a plan. The scientist can assessment the plan and monitor Biomni’s exercise every step of the way in which, because it analyzes information or runs an experiment, intervening at any time to course-correct.

“Biomni is a real partner in biomedical research,” says Jure Leskovec, professor of laptop science at Stanford and Biomni co-creator. “You give the agent a task, and it writes Python code to use the advanced models and tools. From there, the assignment becomes a digital conversation that’s fully documented and auditable. Scientists no longer have to worry about losing the history of their work in notebooks and Excel spreadsheets.”

Jure Leskovec, Biomni co-creator and a Stanford professor of laptop scientist.

High Performance Across Complex Tasks

In Biomni’s first 9 months as an open-source Stanford mission, greater than 15,000 scientists requested the AI analysis assistant to automate 100,000 totally different scientific workflows, comparable to formulating testable hypotheses, performing advanced bioinformatics analyses, and designing rigorous experimental protocols. 

In its preliminary testing, Biomni excelled on established Q&A benchmarks for biomedical data and reasoning. It additionally carried out properly on eight difficult, reasonable eventualities by no means encountered throughout growth, indicating it may generalize throughout domains with out task-specific coaching.

Biomni’s creators spotlight a number of case research to show its potential. In the primary state of affairs, a researcher instructed Biomni to investigate 458 Excel recordsdata containing information from 30 individuals who wore steady glucose displays for a number of months. The researcher then posed an open-ended query: “Can we uncover biologically meaningful thermogenic patterns?” Biomni autonomously generated and executed a 10-step evaluation plan. It inferred meal occasions from glucose spikes and extracted pre- and post-meal physique temperature readings, presenting the ends in a structured, readable report that detailed particular person information and teased out population-level developments.

In different case research, Biomni quickly analyzed large uncooked datasets containing genomic sequences to generate novel insights and designed laboratory protocols to help wet-lab researchers.

“Across each of these use cases, Biomni accelerated the path from messy real-world data to testable hypotheses, and supported applications in domains as diverse as metabolic research and precision health,” Huang says.

Although Biomni approaches human-level efficiency in some duties, comparable to database querying, sequence evaluation, and molecular cloning, the students observe it nonetheless struggles in areas that require nuanced medical judgment, novel experimental causes, or deep organic considering and synthesis. It additionally doesn’t cowl each subject.

Even so, Biomni has set the stage for an period when digital AI biologists will work alongside human researchers to speed up biomedical discovery. In September 2025, the crew spun Biomni out of the Stanford AI Lab and into the industrial world, with a seed spherical of enterprise funding. Huang now leads the startup, referred to as Phylo, whereas Leskovec stays concerned as scientific co-founder. The authentic public platform has migrated to the brand new entity as Biomni Lab, with an Academic Lab Program out there to universities. The codebase stays totally open supply.

“With Biomni, we envision a future where virtual AI biologists operate alongside and augment human scientists to dramatically enhance research productivity, clinical insight, and healthcare,” says Leskovec. “It’s an exciting time for scientific discovery.”

Read extra:

Leskovec is the inaugural Alfred and Rebecca Lin Professor within the School of Engineering.