Unified-MAS: Universally Generating Domain-Specific Nodes for Empowering Automatic Multi-Agent Systems

2026-03-22

AI Summary

Existing automated multi-agent systems (MAS) in knowledge-intensive domains are hindered by the lack of domain-specific expertise in generic nodes and performance degradation caused by architectural coupling during dynamic node generation. This work proposes an offline node synthesis approach that, for the first time, decouples node generation from topological orchestration. In the first phase, domain-specific node blueprints are constructed through external knowledge retrieval and search; in the second phase, a perplexity-guided reinforcement optimization iteratively refines the logical reasoning of bottleneck nodes. By circumventing the intrinsic knowledge limitations of large language models, the method achieves up to a 14.2% performance gain when integrated with Unified-MAS across four specialized domains—including healthcare and law—while significantly reducing computational costs. Furthermore, it demonstrates strong robustness and generalization across diverse models and mathematical reasoning tasks.

Technology Category

• Multiagent Systems: Multiagent Planning
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Automatic Multi-Agent Systems (MAS) generation has emerged as a promising paradigm for solving complex reasoning tasks. However, existing frameworks are fundamentally bottlenecked when applied to knowledge-intensive domains (e.g., healthcare and law). They either rely on a static library of general nodes like Chain-of-Thought, which lack specialized expertise, or attempt to generate nodes on the fly. In the latter case, the orchestrator is not only bound by its internal knowledge limits but must also simultaneously generate domain-specific logic and optimize high-level topology, leading to a severe architectural coupling that degrades overall system efficacy. To bridge this gap, we propose Unified-MAS that decouples granular node implementation from topological orchestration via offline node synthesis. Unified-MAS operates in two stages: (1) Search-Based Node Generation retrieves external open-world knowledge to synthesize specialized node blueprints, overcoming the internal knowledge limits of LLMs; and (2) Reward-Based Node Optimization utilizes a perplexity-guided reward to iteratively enhance the internal logic of bottleneck nodes. Extensive experiments across four specialized domains demonstrate that integrating Unified-MAS into four Automatic-MAS baselines yields a better performance-cost trade-off, achieving up to a 14.2% gain while significantly reducing costs. Further analysis reveals its robustness across different designer LLMs and its effectiveness on conventional tasks such as mathematical reasoning.

Problem

• Automatic Multi-Agent Systems
• knowledge-intensive domains
• domain-specific nodes
• architectural coupling
• node generation

Innovation

• Automatic Multi-Agent Systems
• Offline Node Synthesis
• Search-Based Node Generation
• Reward-Based Optimization
• Domain-Specific Nodes