@article{Gottweis:2026:10.1038/s41586-026-10644-y,
author = {Gottweis, J and Weng, W-H and Daryin, A and Tu, T and Sirkovic, P and Myaskovsky, A and Glowaty, G and Weissenberger, F and Orlandi, A and Popovici, D and Palepu, A and Rong, K and Tanno, R and Saab, K and Zhang, F and Blum, J and Carroll, A and Kulkarni, K and Tomaev, N and Zverinski, D and Rendulic, I and Vedadi, E and Hasler, F and Rimanic, L and Boia, M and Budiselic, I and Feinstein, B and Bellaiche, M and Sheffer, T and Freyberg, J and Ratcliff, J and Bertolli, O and Chou, K and Hassidim, A and Gokturk, B and Vahdat, A and Guan, Y and Dhillon, V and Vaishnav, ED and Lee, B and Costa, TRD and Penadés, JR and Peltz, G and Matias, Y and Manyika, J and Hassabis, D and Xu, Y and Kohli, P and Pawlosky, A and Karthikesalingam, A and Natarajan, V},
doi = {10.1038/s41586-026-10644-y},
journal = {Nature},
title = {Accelerating scientific discovery with Co-Scientist},
url = {http://dx.doi.org/10.1038/s41586-026-10644-y},
year = {2026}
}

Download

TY  - JOUR
AB  - Scientific discovery is driven by scientists generating hypotheses for complex problems that undergo rigorous experimental validation. To augment this process, we introduce Co-Scientist, a multi-agent artificial intelligence (AI) system built on Gemini for structured scientific thinking and hypothesis generation. Co-Scientist aims to help scientists discover new original knowledge. Conditioned on their research objectives and previous scientific evidence, it formulates demonstrably novel research hypotheses for experimental verification. The system’s design involves agents continuously generating, critiquing and refining hypotheses accelerated by scaling test-time compute. Key contributions include (1) a multi-agent architecture with an asynchronous task execution framework for flexible compute scaling, and (2) a tournament evolution process for self-improving hypotheses generation. Automated evaluations show continued benefits of test-time compute scaling, improving hypothesis quality over time. Although this is a general-purpose system, we focus the validation in three biomedical applications: drug repurposing; novel-target discovery1; and explaining mechanisms of antimicrobial resistance2. Specifically, Co-Scientist helped to identify new drug-repurposing candidates and synergistic combination therapies for acute myeloid leukaemia that were validated through in vitro experiments. These real-world validations demonstrate the potential of Co-Scientist to accelerate scientific discovery and usher in an era of AI-empowered scientists.
AU  - Gottweis,J
AU  - Weng,W-H
AU  - Daryin,A
AU  - Tu,T
AU  - Sirkovic,P
AU  - Myaskovsky,A
AU  - Glowaty,G
AU  - Weissenberger,F
AU  - Orlandi,A
AU  - Popovici,D
AU  - Palepu,A
AU  - Rong,K
AU  - Tanno,R
AU  - Saab,K
AU  - Zhang,F
AU  - Blum,J
AU  - Carroll,A
AU  - Kulkarni,K
AU  - Tomaev,N
AU  - Zverinski,D
AU  - Rendulic,I
AU  - Vedadi,E
AU  - Hasler,F
AU  - Rimanic,L
AU  - Boia,M
AU  - Budiselic,I
AU  - Feinstein,B
AU  - Bellaiche,M
AU  - Sheffer,T
AU  - Freyberg,J
AU  - Ratcliff,J
AU  - Bertolli,O
AU  - Chou,K
AU  - Hassidim,A
AU  - Gokturk,B
AU  - Vahdat,A
AU  - Guan,Y
AU  - Dhillon,V
AU  - Vaishnav,ED
AU  - Lee,B
AU  - Costa,TRD
AU  - Penadés,JR
AU  - Peltz,G
AU  - Matias,Y
AU  - Manyika,J
AU  - Hassabis,D
AU  - Xu,Y
AU  - Kohli,P
AU  - Pawlosky,A
AU  - Karthikesalingam,A
AU  - Natarajan,V
DO  - 10.1038/s41586-026-10644-y
PY  - 2026///
SN  - 0028-0836
TI  - Accelerating scientific discovery with Co-Scientist
T2  - Nature
UR  - http://dx.doi.org/10.1038/s41586-026-10644-y
UR  - https://doi.org/10.1038/s41586-026-10644-y
ER  - 

Download