Stanford Researchers Create AI-Driven 'Virtual Scientists' to Accelerate Scientific Discovery

Researchers create 'virtual scientists' to solve complex biological problems

An AI lab developed by Stanford Medicine researchers has already shown promising results, generating ideas for a more effective COVID-19 vaccine in just a few days. There may be a new artificial intelligence-driven tool to turbocharge scientific discovery: virtual labs. Modeled after a well-established Stanford School of Medicine research group, the virtual lab is complete with an AI principal investigator and seasoned scientists. "Good science happens when we have deep, interdisciplinary collaborations where people from different backgrounds work together, and often that's one of the main bottlenecks and challenging parts of research," said James Zou, PhD, associate professor of biomedical data science who led a study detailing the development of the virtual lab. "In parallel, we've seen this tremendous advance in AI agents, which, in a nutshell, are AI systems based on language models that are able to take more proactive actions." People often think of large language models, the type of AI harnessed in this study, as simple question-and-answer bots. "But these are systems that can retrieve data, use different tools, and communicate with each other and with us through human language," Zou said. (The collaboration shown through these AI models is an example of agentic or agential AI, a structure of AI systems that work together to solve complex problems.) The leap in capability gave Zou the idea to start training these models to mimic top-tier scientists in the same way that they think critically about a problem, research certain questions, pose different solutions based on a given area of expertise and bounce ideas off one another to develop a hypothesis worth testing. "There's no shortage of challenges for the world's scientists to solve," said Zou. "The virtual lab could help expedite the development of solutions for a variety of problems." Already, Zou's team has been able to demonstrate the AI lab's potential after tasking the "team" to devise a better way to create a vaccine for SARS-CoV-2, the virus that causes COVID-19. And it took the AI lab only a few days. A paper describing the findings of the study was published July 29 in Nature. Zou and John Pak, PhD, a scientist at Chan Zuckerberg Biohub, are the senior authors of the paper. Kyle Swanson, a computer science graduate student at Stanford University, is the lead author. The virtual lab begins a research project just like any other human lab - with a problem to solve, presented by the lab's leader. The human researcher gives the AI principal investigator, or AI PI, a scientific challenge, and the AI PI takes it from there. "It's the AI PI's job to figure out the other agents and expertise needed to tackle the project," Zou said. For the SARS-CoV-2 project, for instance, the PI agent created an immunology agent, a computation biology agent, and a machine learning agent. And, in every project, no matter the topic, there's one agent that assumes the role of critic. Its job is to poke holes, caution against common pitfalls, and provide constructive criticism to other agents. Zou and his team equipped the virtual scientists with tools and software systems, such as the protein modeling AI system AlphaFold, to better stimulate creative "thinking" skills. The agents even created their own wish list. "They would ask for access to certain tools, and we'd build it into the model to let them use it," Zou said. As research labs go, the virtual team runs a swift operation. Just like Zou's research group, the virtual lab has regular meetings during which agents generate ideas and engage in a conversational back-and-forth. They also have one-on-one meetings, allowing lab members to meet with the PI agent individually to discuss ideas. But unlike human meetings, these virtual gatherings take a few seconds or minutes. On top of that, AI scientists don't get tired, and they don't need snacks or bathroom breaks, so multiple meetings run in parallel. "By the time I've had my morning coffee, they've already had hundreds of research discussions," Zou said during the RAISE Health Symposium, during which he presented on this work. Moreover, the virtual lab is an independent operation. Aside from the initial prompt, the main guideline consistently given to the AI lab members is budget-related, barring any extravagant or outlandish ideas that aren't feasible to validate in the physical lab. Not one prone to micromanagement - in the real or virtual world - Zou estimates that he or his lab members intervene about 1% of the time. "I don't want to tell the AI scientists exactly how they should do their work. That really limits their creativity," Zou said. "I want them to come up with new solutions and ideas that are beyond what I would think about." But that doesn't mean they're not keeping a close eye on what's going on - each meeting, exchange, and interaction in the virtual lab is captured via a transcript, allowing human researchers to track progress and redirect the project if needed. Zou's team put the virtual lab to the test by asking it to devise a new basis for a vaccine against recent COVID-19 variants. Instead of opting for the tried-and-true antibody (a molecule that recognizes and attaches to a foreign substance in the body), the AI team opted for a more unorthodox approach: nanobodies, a fragment of an antibody that's smaller and simpler. "From the beginning of their meetings the AI scientists decided that nanobodies would be a more promising strategy than antibodies - and they provided explanations. They said nanobodies are typically much smaller than antibodies, so that makes the machine learning scientist's job much easier, because when you computationally model proteins, working with smaller molecules means you can have more confidence in modeling and designing them," Zou said. So far, it seems like the AI team is onto something. Pak's team took the nanobody structural designs from the AI researchers and created them in his real-world lab. Not only did they find that the nanobody was experimentally feasible and stable, they also tested its ability to bind to one of the new SARS-CoV-2 variants - a key factor in determining the effectiveness of a new vaccine - and saw that it clung tightly to the virus, more so than existing antibodies designed in the lab. They also measured off-target effects, or whether the nanobody errantly binds to something other than the targeted virus, and found it didn't stray from the COVID-19 spike protein. "The other thing that's promising about these nanobodies is that, in addition to binding well to the recent COVID strain, they're also good at binding to the original strain from Wuhan from five years ago," Zou said, referring to the nanobody's potential to ground a broadly effective vaccine. Now, Zou and his team are analyzing the nanobody's ability to help create a new vaccine. And as they do, they're feeding the experimental data back to the AI lab to further hone the molecular designs. The research team is eager to apply the virtual lab to other scientific questions, and they've recently developed agents that act as sophisticated data analysts that can reassess previously published papers. "The datasets that we collect in biology and medicine are very complex, and we're just scratching the surface when we analyze those data," Zou said. "Often the AI agents are able to come up with new findings beyond what the previous human researchers published on. I think that's really exciting."

Researchers create 'virtual scientists' to solve complex biological problems

There may be a new artificial intelligence-driven tool to turbocharge scientific discovery: virtual labs. Modeled after a well-established Stanford School of Medicine research group, the virtual lab is complete with an AI principal investigator and seasoned scientists. "Good science happens when we have deep, interdisciplinary collaborations where people from different backgrounds work together, and often that's one of the main bottlenecks and challenging parts of research," said James Zou, Ph.D., associate professor of biomedical data science who led a study detailing the development of the virtual lab. "In parallel, we've seen this tremendous advance in AI agents, which, in a nutshell, are AI systems based on language models that are able to take more proactive actions." People often think of large language models, the type of AI harnessed in this study, as simple question-and-answer bots. "But these are systems that can retrieve data, use different tools, and communicate with each other and with us through human language," Zou said. (The collaboration shown through these AI models is an example of agentic or agential AI, a structure of AI systems that work together to solve complex problems.) The leap in capability gave Zou the idea to start training these models to mimic top-tier scientists in the same way that they think critically about a problem, research certain questions, pose different solutions based on a given area of expertise and bounce ideas off one another to develop a hypothesis worth testing. "There's no shortage of challenges for the world's scientists to solve," said Zou. "The virtual lab could help expedite the development of solutions for a variety of problems." Already, Zou's team has been able to demonstrate the AI lab's potential after tasking the "team" to devise a better way to create a vaccine for SARS-CoV-2, the virus that causes COVID-19. And it took the AI lab only a few days. A paper describing the findings of the study was published in Nature. Zou and John Pak, Ph.D., a scientist at Chan Zuckerberg Biohub, are the senior authors of the paper. Kyle Swanson, a computer science graduate student at Stanford University, is the lead author. Running a virtual lab The virtual lab begins a research project just like any other human lab -- with a problem to solve, presented by the lab's leader. The human researcher gives the AI principal investigator, or AI PI, a scientific challenge, and the AI PI takes it from there. "It's the AI PI's job to figure out the other agents and expertise needed to tackle the project," Zou said. For the SARS-CoV-2 project, for instance, the PI agent created an immunology agent, a computation biology agent and a machine learning agent. And, in every project, no matter the topic, there's one agent that assumes the role of critic. Its job is to poke holes, caution against common pitfalls and provide constructive criticism to other agents. Zou and his team equipped the virtual scientists with tools and software systems, such as the protein modeling AI system AlphaFold, to better stimulate creative "thinking" skills. The agents even created their own wish list. "They would ask for access to certain tools, and we'd build it into the model to let them use it," Zou said. As research labs go, the virtual team runs a swift operation. Just like Zou's research group, the virtual lab has regular meetings during which agents generate ideas and engage in a conversational back-and-forth. They also have one-on-one meetings, allowing lab members to meet with the PI agent individually to discuss ideas. But unlike human meetings, these virtual gatherings take a few seconds or minutes. On top of that, AI scientists don't get tired, and they don't need snacks or bathroom breaks, so multiple meetings run in parallel. "By the time I've had my morning coffee, they've already had hundreds of research discussions," Zou said during the RAISE Health Symposium, during which he presented this work. Moreover, the virtual lab is an independent operation. Aside from the initial prompt, the main guideline consistently given to the AI lab members is budget-related, barring any extravagant or outlandish ideas that aren't feasible to validate in the physical lab. Not one prone to micromanagement -- in the real or virtual world -- Zou estimates that he or his lab members intervene about 1% of the time. "I don't want to tell the AI scientists exactly how they should do their work. That really limits their creativity," Zou said. "I want them to come up with new solutions and ideas that are beyond what I would think about." But that doesn't mean they're not keeping a close eye on what's going on -- each meeting, exchange and interaction in the virtual lab is captured via a transcript, allowing human researchers to track progress and redirect the project if needed. SARS-CoV-2 and beyond Zou's team put the virtual lab to the test by asking it to devise a new basis for a vaccine against recent COVID-19 variants. Instead of opting for the tried-and-true antibody (a molecule that recognizes and attaches to a foreign substance in the body), the AI team opted for a more unorthodox approach: nanobodies, a fragment of an antibody that's smaller and simpler. "From the beginning of their meetings, the AI scientists decided that nanobodies would be a more promising strategy than antibodies -- and they provided explanations. They said nanobodies are typically much smaller than antibodies, so that makes the machine learning scientist's job much easier, because when you computationally model proteins, working with smaller molecules means you can have more confidence in modeling and designing them," Zou said. So far, it seems like the AI team is onto something. Pak's team took the nanobody structural designs from the AI researchers and created them in his real-world lab. Not only did they find that the nanobody was experimentally feasible and stable, they also tested its ability to bind to one of the new SARS-CoV-2 variants -- a key factor in determining the effectiveness of a new vaccine -- and saw that it clung tightly to the virus, more so than existing antibodies designed in the lab. They also measured off-target effects, or whether the nanobody errantly binds to something other than the targeted virus, and found it didn't stray from the COVID-19 spike protein. "The other thing that's promising about these nanobodies is that, in addition to binding well to the recent COVID strain, they're also good at binding to the original strain from Wuhan from five years ago," Zou said, referring to the nanobody's potential to ground a broadly effective vaccine. Now, Zou and his team are analyzing the nanobody's ability to help create a new vaccine. And as they do, they're feeding the experimental data back to the AI lab to further hone the molecular designs. The research team is eager to apply the virtual lab to other scientific questions, and they've recently developed agents that act as sophisticated data analysts that can reassess previously published papers. "The datasets that we collect in biology and medicine are very complex, and we're just scratching the surface when we analyze those data," Zou said. "Often the AI agents are able to come up with new findings beyond what the previous human researchers published on. I think that's really exciting."

With no need for sleep or food, AI-built 'scientists' quickly design nanobodies against SARS-CoV-2 variants

Imagine you're a molecular biologist wanting to launch a project seeking treatments for a newly emerging disease. You know you need the expertise of a virologist and an immunologist, plus a bioinformatics specialist to help analyze and generate insights from your data. But you lack the resources or connections to build a big multidisciplinary team. Researchers at Chan Zuckerberg Biohub San Francisco and Stanford University now offer a novel solution to this dilemma: an AI-driven Virtual Lab through which a team of AI agents, each equipped with varied scientific expertise, can tackle sophisticated and open-ended scientific problems by formulating, refining, and carrying out a complex research strategy -- these agents can even conduct virtual experiments, producing results that can be validated in real-life laboratories. In a study published in Nature on July 29, 2025, co-senior authors John Pak of CZ Biohub SF and Stanford's James Zou describe their Virtual Lab platform, in which a human user creates a "Principal Investigator" AI agent (the PI) that assembles and directs a team of additional AI agents emulating the specialized research roles seen in science labs. The human researcher proposes a scientific question, and then monitors meetings in which the PI agent exchanges ideas with the team of specialist agents to advance the research. The agents are run by a large language model (LLM), giving them scientific reasoning and decision-making capabilities. The authors then used the Virtual Lab platform to investigate a timely research question: designing antibodies or nanobodies to bind to the spike protein of new variants of the SARS-CoV-2 virus. After just a few days of working together, the Virtual Lab team had designed and implemented an innovative computational pipeline, and had presented Pak and Zou with blueprints for dozens of binders, two of which showed particular promise against new SARS-CoV-2 strains when subsequently tested in Pak's lab. The overall Virtual Lab study was led by Kyle Swanson, a Ph.D. student in Zou's group. "What was once this crazy science fiction idea is now a reality," said Pak, group leader of the Biohub SF Protein Sciences Platform. "The AI agents came up with a pipeline that was quite creative. But at the same time, it wasn't outrageous or nonsensical. It was very reasonable -- and they were very fast." Zou is a pioneering AI researcher who has been recognized widely for breakthroughs in using AI for biomedical research, including winning the International Society of Computational Biology's 2025 Overton Prize and being named in the New York Times' 2024 Good Tech Awards for SyntheMol, an AI system that can design and validate novel antibiotics. "This is the first demonstration of autonomous AI agents really solving a challenging research problem, from start to finish," said Zou, an associate professor of biomedical data science who leads Stanford University's AI for Health program and is also a CZ Biohub SF Investigator. "The AI agents made good decisions about complex problems and were able to quickly design dozens of protein candidates that we could then test in lab experiments." A fortuitous real-world meeting It's become increasingly common for human scientists to employ LLMs to help with science research, such as analyzing data, writing code, and even designing proteins. Zou and Pak's Virtual Lab platform, however, is to their knowledge the first to apply multistep reasoning and interdisciplinary expertise to successfully address an open-ended research question. Zou and Pak first met at one of the biweekly Biohub SF Investigator Program meetings. "I had just seen James give a talk at the previous Investigator meeting, where he said he wished he could do more experimental work," Pak said. "So I decided to introduce myself." That conversation, in the spring of 2024, sparked a collaboration that drew on Zou's AI expertise and Pak's expertise in protein science. In addition to the PI agent and specialist agents, their Virtual Lab platform includes a Scientific Critic agent, a generalist whose role is to ask probing questions and inject a dose of skepticism into the process. "We found the Critic to be quite essential, and also reduced hallucinations," Zou said. While human researchers participated in AI scientists' meetings and offered guidance at key moments, their words made up only about 1% of all conversations. The vast majority of discussions, decisions, and analyses were performed by the AI agents themselves. In this study, the Virtual Lab team came up with 92 new "nanobodies" (tiny proteins that work like antibodies), and experiments in Pak's lab found that two bound to the so-called spike protein of recent SARS-CoV-2 variants, a significant enough finding that Pak expects to publish studies on them. "You'd think there'd be no way AI agents talking together could propose something akin to what a human scientist would come up with, but we found here that they really can," said Pak. "It's pretty shocking." When asked if he's worried about AI scientists replacing him, Pak says no. Instead, he thinks these new virtual collaborators will just enhance his work. "This project opened the door for our Protein Science team to test a lot more well-conceived ideas very quickly," he said. "The Virtual Lab gave us more work, in a sense, because it gave us more ideas to test. If AI can produce more testable hypotheses, that's more work for everyone." The results, said Pak and Zou, not only demonstrate the potential benefits of human-AI collaborations but also highlight the importance of diverse perspectives in science. Even in these virtual settings, instructing agents to assume different roles and bring varying perspectives to the table resulted in better outcomes than one AI agent working alone, they said. And because the discussions result in a transcript that human team members can access and review, researchers can feel confident about why certain decisions were made and probe further if they have questions or concerns. "The agents and the humans are all speaking the same language, so there's nothing 'black box' about it, and the collaboration can progress very smoothly," Pak said. "It was a really positive experience overall, and I feel pretty confident about applying the Virtual Lab in future research." Zou says the existing platform is designed for biomedical research questions, but modifications would allow it to be used in a much wider array of scientific disciplines. "We're demonstrating a new paradigm where AI is not just a tool we use for a specific step in our research, but it can actually be a primary driver of the whole process to generate discoveries," said Zou. "It's a big shift, and we're excited to see how it helps us advance in all areas of research."

With No Need for Sleep or Food, AI-Built 'Scientists' Get the Job Done Quickly | Newswise

Newswise -- Imagine you're a molecular biologist wanting to launch a project seeking treatments for a newly emerging disease. You know you need the expertise of a virologist and an immunologist, plus a bioinformatics specialist to help analyze and generate insights from your data. But you lack the resources or connections to build a big multidisciplinary team. Researchers at Chan Zuckerberg Biohub San Francisco and Stanford University now offer a novel solution to this dilemma: an AI-driven Virtual Lab through which a team of AI agents, each equipped with varied scientific expertise, can tackle sophisticated and open-ended scientific problems by formulating, refining, and carrying out a complex research strategy -- these agents can even conduct virtual experiments, producing results that can be validated in real-life laboratories. In a study published in Nature on July 29, 2025, co-senior authors John Pak of CZ Biohub SF and Stanford's James Zou describe their Virtual Lab platform, in which a human user creates a "Principal Investigator" AI agent (the PI) that assembles and directs a team of additional AI agents emulating the specialized research roles seen in science labs. The human researcher proposes a scientific question, and then monitors meetings in which the PI agent exchanges ideas with the team of specialist agents to advance the research. The agents are run by a large language model (LLM), giving them scientific reasoning and decision-making capabilities. The authors then used the Virtual Lab platform to investigate a timely research question: designing antibodies or Nanobodies to bind to the spike protein of new variants of the SARS-CoV-2 virus. After just a few days working together, the Virtual Lab team had designed and implemented an innovative computational pipeline, and had presented Pak and Zou with blueprints for dozens of binders, two of which showed particular promise against new SARS-CoV-2 strains when subsequently tested in Pak's lab. The overall Virtual Lab study was led by Kyle Swanson, a Ph.D. student in Zou's group. "What was once this crazy science fiction idea is now a reality," said Pak, group leader of the Biohub SF Protein Sciences Platform. "The AI agents came up with a pipeline that was quite creative. But at the same time, it wasn't outrageous or nonsensical. It was very reasonable - and they were very fast." Zou is a pioneering AI researcher who has been recognized widely for breakthroughs in using AI for biomedical research, including winning the International Society of Computational Biology's 2025 Overton Prize and being named in the New York Times' 2024 Good Tech Awards for SyntheMol, an AI system that can design and validate novel antibiotics. "This is the first demonstration of autonomous AI agents really solving a challenging research problem, from start to finish," said Zou, an associate professor of biomedical data science who leads Stanford University's AI for Health program and is also a CZ Biohub SF Investigator. "The AI agents made good decisions about complex problems and were able to quickly design dozens of protein candidates that we could then test in lab experiments." A fortuitous real-world meeting It's become increasingly common for human scientists to employ LLMs to help with science research, such as analyzing data, writing code, and even designing proteins. Zou and Pak's Virtual Lab platform, however, is to their knowledge the first to apply multistep reasoning and interdisciplinary expertise to successfully address an open-ended research question. Zou and Pak first met at one of the biweekly Biohub SF Investigator Program meetings. "I had just seen James give a talk at the previous Investigator meeting, where he said he wished he could do more experimental work," Pak said. "So I decided to introduce myself." That conversation, in the spring of 2024, sparked a collaboration that drew on Zou's AI expertise and Pak's expertise in protein science. In addition to the PI agent and specialist agents, their Virtual Lab platform includes a Scientific Critic agent, a generalist whose role is to ask probing questions and inject a dose of skepticism into the process. "We found the Critic to be quite essential, and also reduced hallucinations," Zou said. While human researchers participated in AI scientists' meetings and offered guidance at key moments, their words made up only about 1% of all conversations. The vast majority of discussions, decisions, and analyses were performed by the AI agents themselves. In this study, the Virtual Lab team came up with 92 new "Nanobodies" (tiny proteins that work like antibodies), and experiments in Pak's lab found that two bound to the so-called spike protein of recent SARS-CoV-2 variants, a significant enough finding that Pak expects to publish studies on them. "You'd think there'd be no way AI agents talking together could propose something akin to what a human scientist would come up with, but we found here that they really can," said Pak. "It's pretty shocking." When asked if he's worried about AI scientists replacing him, Pak says no. Instead, he thinks these new virtual collaborators will just enhance his work. "This project opened the door for our Protein Science team to test a lot more well-conceived ideas very quickly," he said. "The Virtual Lab gave us more work, in a sense, because it gave us more ideas to test. If AI can produce more testable hypotheses, that's more work for everyone." The results, said Pak and Zou, not only demonstrate the potential benefits of human-AI collaborations but also highlight the importance of diverse perspectives in science. Even in these virtual settings, instructing agents to assume different roles and bring varying perspectives to the table resulted in better outcomes than one AI agent working alone, they said. And because the discussions result in a transcript that human team members can access and review, researchers can feel confident about why certain decisions were made and probe further if they have questions or concerns. "The agents and the humans are all speaking the same language, so there's nothing 'black box' about it, and the collaboration can progress very smoothly," Pak said. "It was a really positive experience overall, and I feel pretty confident about applying the Virtual Lab in future research." Zou says the existing platform is designed for biomedical research questions, but modifications would allow it to be used in a much wider array of scientific disciplines. "We're demonstrating a new paradigm where AI is not just a tool we use for a specific step in our research, but it can actually be a primary driver of the whole process to generate discoveries," said Zou. "It's a big shift, and we're excited to see how it helps us advance in all areas of research." # # # About CZ Biohub San Francisco: A nonprofit biomedical research center founded in 2016, CZ Biohub SF is part of the CZ Biohub Network, a group of research institutes created and supported by the Chan Zuckerberg Initiative. CZ Biohub SF's researchers, engineers, and data scientists, in collaboration with colleagues at our partner universities -- Stanford University; the University of California, Berkeley; and the University of California, San Francisco -- seek to understand the fundamental mechanisms underlying disease and develop new technologies that will lead to actionable diagnostics and effective therapies. Learn more at czbiohub.org/sf.