【Proposal】GoalAct: A Globally Adaptive Dynamic Legal Multi-agent Collaboration System

Junjie Chen, Ruowen Zhao, ZhiYuan Feng

Keywords: Multi-agent, Tool Learning, Legal Assistant, Planning, Reflection, Memory

Abstract: This paper presents our proposed multi-agent collaboration system based on GLM-4, which employs a strategy that combines global and local information to provide legal services by accessing relevant legal databases through API calls. The strength of our approach lies in integrating planning, reflection and memory globally and locally, thereby enhancing both the accuracy and adaptability.

Submission Number: 4