GBC: Gradient-Based Connections for Optimizing Multi-Agent Systems
2026-06-26 • Multiagent Systems
Multiagent Systems
AI summaryⓘ
The authors studied systems where multiple AI agents, based on large language models, work together to solve tasks. They found that it's hard to know which agent or step causes errors because feedback is usually too broad. They created a method called Gradient-Based Connections (GBC) that tracks how each agent's output affects others at a very detailed level. This helps identify and fix mistakes more precisely. Their implementation, AgentChord, showed better performance in tests compared to other methods.
Multi-agent systemsLarge language modelsCredit assignmentGradient-based optimizationComputational graphToken-level attributionPrompt optimizationMultiWOZPrefix gradient computationError attribution
Authors
Xiaocheng Yang, Abdulrahman Alrabah, Dilek Hakkani-Tür, Gokhan Tur
Abstract
Multi-agent systems (MAS) built on large language models (LLMs) provide a promising framework for solving complex tasks through role specialization and structured interaction. However, their performance is often limited by miscoordination and, more fundamentally, the lack of fine-grained credit assignment across agents. Existing approaches typically rely on coarse-grained feedback, making it difficult to identify which agents or interaction steps are responsible for errors. We propose Gradient-Based Connections (GBC), an approach for fine-grained attribution and optimization of multi-agent systems. GBC models a MAS as a computational graph and introduces gradient-based connection weights to quantify the influence of each agent's output on downstream agents at the token level. By constructing an attribution graph and propagating task-specific loss signals backward, our method enables precise identification of error sources and targeted prompt optimization. We further develop AgentChord, an efficient implementation that leverages prefix-based gradient computation. Experiments on MultiWOZ and τ-bench show that GBC improves multi-agent performance and outperforms strong single-agent and multi-agent baselines, and higher attribution quality is associated with greater optimization effectiveness. Code is available at: https://github.com/yxc-cyber/AgentChord.