Launched this week, this is Google’s new AI co-scientist system, which has shown remarkable early results, solving a decade-old antibiotic resistance puzzle in just… two days. The system’s power comes from its innovative multi-agent approach – six specialised AI agents working together, each handling different aspects of the scientific process from generating hypotheses to reflecting on results. Like a well-coordinated research team, these agents collaborate under a supervisor that manages the workflow. Scientists simply input their research goals, and the system orchestrates the agents to explore, analyse and generate detailed hypotheses. Early tests at Imperial College London suggest this structured, team-based AI approach could transform how scientists tackle complex research challenges.