The automated synthesis of plasmonic nanoparticles with on-demand properties is a challenging task. Here the authors integrate a fluidic reactor, real-time characterization, and machine learning in a self-driven lab for the photochemical synthesis of nanoparticles with targeted properties.