A multi-agent AI system described as Robin demonstrated an end-to-end workflow for experimental biology, including hypothesis generation, experiment design, and data analysis—then closed the loop with updated hypotheses. The research, published in Nature, reports the system automated key steps used to identify therapeutic candidates for dry age-related macular degeneration. The study describes Robin proposing a phagocytosis-enhancement strategy, then validating candidate drugs in vitro and supporting mechanism exploration with follow-up RNA-seq analysis. The work positions the system as semi-autonomous “lab-in-the-loop” discovery, rather than a tool limited to analysis. For biotech R&D leaders, the immediate implication is acceleration of iteration cycles—moving from slow, human-driven hypothesis-to-experiment conversion toward more continuous automated discovery scaffolding.