LLM-Driven Autonomous Agents for Chemical Synthesis and Discovery

ABSTRACT

The discovery of novel chemical compounds is traditionally hindered by the slow, iterative process of manual hypothesis generation and bench testing. We present ChemAgent, an Agentic AI framework that autonomously designs synthetic pathways, interfaces with cloud-based robotic laboratories via APIs, and executes chemical syntheses. By employing a multi-agent architecture featuring a 'Literature Analyst', a 'Synthesis Planner', and a 'Safety Validator', ChemAgent successfully synthesized four novel organic compounds in a continuous 72-hour automated loop, marking a paradigm shift in autonomous scientific discovery.