AI co-scientist is a multi-agent artificial intelligence system built on the Gemini 2.0 model that serves as a virtual collaborator for scientists to generate new hypotheses and research proposals.[1] The system was tested in three biomedical applications: drug repurposing, designing new therapeutic targets, and elucidating mechanisms of antibiotic resistance.[1] In the area of drug repositioning, the system suggested existing drugs for the treatment of acute myeloid leukemia, with laboratory tests confirming that Kira 6 dose-dependently reduced the growth of cancer cells.[3] In identifying targets for the treatment of liver fibrosis, the system suggested epigenetic targets with significant antifibrotic activity in human hepatic organoids.[1] All proposed treatments showed promising activity with p-values below 0.01 compared to controls.[1] The hypotheses of the system were verified in real laboratory experiments with the guidance of experts.[1] These results accelerate biomedical discoveries and complement human creativity.[1] Details of liver fibrosis will be presented in a report by collaborators from Stanford University.[1]