Decoupled Multi-Agent Orchestration

2026-10-05

AI Summary

This study addresses the poor transferability and ambiguous credit assignment arising from the tight coupling between existing planning methods and fixed worker pools by proposing DeOrch. This framework introduces a novel task-agnostic, two-stage planning mechanism that decouples task decomposition from collaborative decision-making. Furthermore, it incorporates a lightweight matcher to evaluate action compatibility and employs contextual bandits for online adaptation to new workers, enabling cross-pool transfer without retraining. Experimental results demonstrate that DeOrch consistently outperforms existing methods on both in-distribution and out-of-distribution tasks, significantly reducing invocation counts while maintaining strong generalization capabilities across unseen worker pools.

Abstract

Learned orchestration can automatically construct effective language-model multi-agent systems, but existing approaches couple planning to fixed worker pools and train decomposition and collaboration from the same terminal outcome, limiting transfer and obscuring credit assignment. We introduce DeOrch, which separates worker-agnostic planning from concrete worker selection. Its two-stage planner first decomposes the task without worker information, then chooses collaboration operations using compact, worker-identity-free matchability feedback from the pool, enabling conditional credit assignment to decomposition and collaboration decisions. A lightweight matcher estimates worker suitability from behavior on a fixed probe set and adapts online with a contextual bandit, allowing new workers to be incorporated without retraining the planner or matcher. Across diverse in- and out-of-distribution tasks, DeOrch outperforms prior automatic MAS orchestration methods with fewer worker calls than competing learned orchestrators, remains effective when transferred to an entirely unseen worker pool without retraining, and shows consistent gains from both components.

Problem

• multi-agent orchestration
• decoupled planning
• credit assignment
• transferability
• worker-agnostic

Innovation

• Decoupled Multi-Agent Orchestration
• Credit Assignment
• Contextual Bandit
• Worker-Agnostic Planning
• Transferability