JoyAgent-JDGenie introduces multi-agent architecture with hierarchical memory for GAIA benchmark

JoyAgent-JDGenie is a technical report proposing a generalist agent architecture designed to address the lack of unifying design for robustness and adaptability in existing autonomous systems. The framework integrates three core components: a collective multi-agent system combining planning and execution agents with critic model voting, a hierarchical memory system spanning working, semantic, and procedural layers, and a refined tool suite for search, code execution, and multimodal parsing.

• The architecture employs a collective multi-agent framework that combines planning and execution agents with critic model voting.
• It utilizes a hierarchical memory system consisting of working, semantic, and procedural layers.
• The system includes a refined tool suite supporting search, code execution, and multimodal parsing.
• Evaluated on the GAIA benchmark, the framework consistently outperforms open-source baselines and approaches the performance of proprietary systems.

These results demonstrate the importance of system-level integration and highlight a path toward scalable, resilient, and adaptive AI assistants capable of operating across diverse domains and tasks.