Multi-Agent Coordination Adaptation via Structure-Guided Orchestration
2026-05-25 • Multiagent Systems
Multiagent SystemsArtificial Intelligence
AI summaryⓘ
The authors looked at how groups of AI agents work together to solve complicated tasks. They found that existing methods either fix the agents' roles in advance, which limits flexibility, or change decisions on the fly but lack clear coordination. To fix this, they created MACA, which learns the best way for agents to join and interact based on the task and resources, using a probabilistic approach. This helps the agents coordinate better, use less computation, and perform tasks more efficiently compared to other methods.
large language modelsmulti-agent systemscoordinationposterior inferencestructural priororchestrationpolicy-based methodstoken efficiencyprobabilistic modelingtask adaptation
Authors
Haoran Li, Shulun Chen, Shaoyuan Sun, Hanchen Wang
Abstract
As large language model (LLM)-based multi-agent systems scale to handle increasingly complex tasks, balancing structural stability and dynamic adaptability becomes increasingly challenging. Existing systems typically adopt either structure-centric methods, committing to structures determined upfront that limit fine-grained control, or orchestration-centric methods, adapting decisions dynamically while leaving coordination structure implicit and unstable. To address this challenge, we revisit multi-agent coordination from a probabilistic perspective, casting it as posterior inference over the joint distribution of structure and orchestration. We introduce MACA, an automated coordination framework that learns a task- and budget-conditioned structural prior over agent participation and interactions. This prior guides a policy-based orchestration as an approximation to posterior inference, enabling efficient solutions with fine-grained control. Across benchmarks, MACA outperforms adaptive multi-agent baselines by an average of 8.42% while using 43.19% fewer tokens. Further investigation reveals that joint adaptation of structure and orchestration suppresses redundant interactions, converging coordination toward task-effective execution.