A multi-agent system for automating scientific discovery
6 affiliations
- 1FutureHouse, San Francisco, CA, USA
- 2University of Oxford, Oxford, UK
- 3Fordham University, New York, NY, USA
- 4FutureHouse, San Francisco, CA, USA. [email protected]
- 5FutureHouse, San Francisco, CA, USA. [email protected]
- 6FutureHouse, San Francisco, CA, USA. [email protected]
Abstract
Scientific discovery is driven by the iterative process of observation, hypothesis generation, experimentation, and data analysis. Despite recent advancements in applying artificial intelligence to biology, no system has yet automated all these stages [1, 2, 3]. Here, we introduce Robin, the first multi-agent system capable of fully automating both hypothesis generation and data analysis for experimental biology. By integrating literature search agents with data analysis agents, Robin can generate hypotheses, propose experiments, interpret experimental results, and generate updated hypotheses, achieving a semi-autonomous approach to scientific discovery. By applying this system, we were able to identify promising therapeutic candidates for dry age-related macular degeneration (dAMD), the major cause of blindness in the developed world [4, 5]. Robin proposed enhancing retinal pigment epithelium phagocytosis as a therapeutic strategy, and identified and confirmed in vitro efficacy for ripasudil and KL001. Ripasudil is a clinically-used Rho kinase (ROCK) inhibitor that has never previously been proposed for treating dAMD. To elucidate the mechanism of ripasudil-induced upregulation of phagocytosis, Robin then proposed and analyzed a follow-up RNA-seq experiment, which revealed upregulation of ABCA1 , a lipid efflux pump and possible novel target. All hypotheses, experimental directions, data analyses, and data figures in the main text of this report were produced by Robin. As the first AI system to autonomously discover and validate novel therapeutic candidates within an iterative lab-in-the-loop framework, Robin establishes a new paradigm for AI-driven scientific discovery.
Related
- Autonomous in-silico inorganic materials discovery via multi-agent physics-aware scientific reasoning
- The Virtual Biotech: A multi-agent AI framework for therapeutic discovery and development
- Hierarchical Multi-agent Large Language Model Reasoning for Autonomous Heterogeneous Catalyst Discovery
- GVIM 2.0: A Multi-Agent Research System for Chemistry and Materials Science
- Accelerating scientific discovery with Co-Scientist
BibTeX
@article{ghareeb2026multiagent,
title = {{A multi-agent system for automating scientific discovery}},
author = {Ali Essa Ghareeb and Benjamin Chang and Ludovico Mitchener and Yiu Angela and Caralyn J. Szostkiewicz and Dmytro Shved and Gavin J. Gyimesi and Jon M. Laurent and Samantha M. Wright and Muhammed T. Razzak and Andrew D. White and Silvia C. Finnemann and Michaela M. Hinks and Samuel G. Rodriques},
journal = {Nature},
year = {2026},
month = may,
doi = {10.1038/s41586-026-10652-y},
url = {https://doi.org/10.1038/s41586-026-10652-y},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - A multi-agent system for automating scientific discovery
AU - Ali Essa Ghareeb
AU - Benjamin Chang
AU - Ludovico Mitchener
AU - Yiu Angela
AU - Caralyn J. Szostkiewicz
AU - Dmytro Shved
AU - Gavin J. Gyimesi
AU - Jon M. Laurent
AU - Samantha M. Wright
AU - Muhammed T. Razzak
AU - Andrew D. White
AU - Silvia C. Finnemann
AU - Michaela M. Hinks
AU - Samuel G. Rodriques
PY - 2026
DA - 2026/05/19
JO - Nature
DO - 10.1038/s41586-026-10652-y
UR - https://doi.org/10.1038/s41586-026-10652-y
AB - Scientific discovery is driven by the iterative process of observation, hypothesis generation, experimentation, and data analysis. Despite recent advancements in applying artificial intelligence to biology, no system has yet automated all these stages [1, 2, 3]. Here, we introduce Robin, the first multi-agent system capable of fully automating both hypothesis generation and data analysis for experimental biology. By integrating literature search agents with data analysis agents, Robin can generate hypotheses, propose experiments, interpret experimental results, and generate updated hypotheses, achieving a semi-autonomous approach to scientific discovery. By applying this system, we were able to identify promising therapeutic candidates for dry age-related macular degeneration (dAMD), the major cause of blindness in the developed world [4, 5]. Robin proposed enhancing retinal pigment epithelium phagocytosis as a therapeutic strategy, and identified and confirmed in vitro efficacy for ripasudil and KL001. Ripasudil is a clinically-used Rho kinase (ROCK) inhibitor that has never previously been proposed for treating dAMD. To elucidate the mechanism of ripasudil-induced upregulation of phagocytosis, Robin then proposed and analyzed a follow-up RNA-seq experiment, which revealed upregulation of ABCA1 , a lipid efflux pump and possible novel target. All hypotheses, experimental directions, data analyses, and data figures in the main text of this report were produced by Robin. As the first AI system to autonomously discover and validate novel therapeutic candidates within an iterative lab-in-the-loop framework, Robin establishes a new paradigm for AI-driven scientific discovery.
SN - 0028-0836
SN - 1476-4687
ER -