Expert-Inspired Multi-Agent Coordination for Multi-Objective Molecular Optimization

Multi-objective molecular optimization is a core challenge in drug discovery because it requires improving several, often conflicting, molecular properties at once. Inspired by the division of labor in medicinal chemistry, this AAAI-26 paper introduces MAMO, a multi-agent framework in which each agent specializes in one objective and a central scheduling module reallocates tasks from evaluation feedback. The coordination mechanism supports interpretable, goal-conditioned optimization and, on benchmark datasets, improves both objective quality and Pareto diversity under strong inter-objective conflict.