A peer-reviewed Nature report describes Robin, a multi-agent system that automates end-to-end experimental biology workflows—spanning hypothesis generation, experiment design, interpretation, and iterative updating. The system combined literature-search agents with data-analysis agents to propose therapeutic directions and then validate them in vitro and via follow-up experiments. In the study, Robin identified candidates for dry age-related macular degeneration (dAMD), proposing retinal pigment epithelium phagocytosis as a therapeutic strategy and confirming in vitro efficacy for the ROCK inhibitor ripasudil and KL001. The workflow then used follow-on RNA-seq to refine mechanistic understanding and connect results to ABCA1 as a potential target. The paper positions Robin as a “lab-in-the-loop” framework that can produce hypotheses, experimental directions, and figures with minimal human steering beyond evaluation. It also reports high concordance with human expert judgment for decisions in the system. For biotech R&D teams, the actionable signal is not that AI replaces biology, but that AI systems are now being validated as iterative experimental planners for specific translational questions.