August 9, 2025
5 min read
PYMNTS
Stanford AI Agents’ Lab Rapidly Discovers Promising COVID-19 Nanobody Drug Leads
Stanford University researchers have developed a virtual laboratory powered by artificial intelligence (AI) agents that rapidly designed novel molecules targeting fast-evolving COVID-19 variants. This innovative approach led to the discovery of two promising nanobody drug candidates within days, demonstrating strong binding to virus strains that evade existing antibody therapies.AI Agents Accelerate Drug Discovery
Professor James Zou of Stanford, who leads the project, sought to overcome the limitations of human researchers’ time by creating AI agents modeled after his lab. These agents autonomously conducted experiments and designed molecules, allowing exploration of hundreds of research ideas simultaneously. "At the time of this project, there were actually no good, known binders," Zou explained. "This is a very challenging, open-research problem, but also a very important and impactful problem from a public health perspective." The AI agents sifted through trillions of molecular combinations to propose 92 novel nanobody candidates. Nanobodies, smaller and more stable cousins of antibodies typically found in animals like camels, were chosen over traditional antibodies due to their computational modeling advantages and potential stability.Novel Nanobodies Show Strong Binding
Physical experiments confirmed that these AI-designed nanobodies bound effectively to the latest coronavirus strains, outperforming existing human-designed antibodies. Two candidates stood out for their strong binding to SARS-CoV-2 variants, including those that evade current antibody therapies, while also protecting against the original strain. "These novel nanobodies created by the virtual lab do show binding," Zou said. "I’m quite impressed by the output of the results."Collaborative AI Agent Framework
The virtual lab operates with two lead AI agents powered by GPT-4o from OpenAI: the Principal Investigator (PI) and the Scientific Critic (SC). The PI manages a team of specialized agents—an immunologist, a computational biologist, and a machine learning specialist—who collaborate through team and individual meetings to design and refine drug candidates. The SC challenges conclusions to minimize hallucinations and improve robustness. Multiple parallel meetings are conducted to synthesize consensus, enhancing reliability. Remarkably, the human researcher contributed only 1% of the work, with the AI agents generating over 120,000 words of research content in a matter of days. "Collaboration between AI and human is certainly much more effective than either the human alone or the AI alone," Zou emphasized. "In problems like medicine and diseases, there’s actually no shortage of problems to solve."Open Source and Broader Applications
The virtual lab platform is fully open source and available on GitHub, enabling researchers worldwide to adapt it for various scientific challenges, including biomarker discovery for Alzheimer’s disease. Zou’s findings are detailed in a Nature paper, co-authored with John Pak of the Chan Zuckerberg Biohub and Stanford computer science graduate Kyle Swanson.AI in Healthcare Drug Discovery
The use of AI in healthcare is gaining momentum, with leaders like Google DeepMind CEO Demis Hassabis predicting AI-designed drugs entering clinical trials soon. Stanford’s virtual lab exemplifies how AI can accelerate drug discovery, especially in urgent public health crises.Source: Originally published at PYMNTS on August 8, 2025.