dlvr.it
Catalyst: When self-driving labs begin to reason
Self-driving laboratories (SDLs) are evolving from automated experimental engines into reasoning scientific partners. In agentic SDLs, artificial intelligence (AI) agents coordinate perception, planning, experimentation, analysis, and reflection across the research cycle. By making metadecisions, including what to measure, which models to trust, when to ask humans, and how to revise hypotheses, agentic SDLs move beyond closed-loop optimization toward adaptive, context-aware scientific discovery. This Catalysis article discusses how reasoning agents can expand autonomy, accelerate learning, and reshape scientific practice.