Bus communication framework improves multi-agent collaboration and reasoning
BusMA: A Bus Communication Substrate for Multi-Agent Systems
Artificial Intelligence
Summary
Complex tasks often require multiple agents to work together, but current systems limit how they communicate. The authors created BusMA, a method inspired by computer buses, where agents share a common channel to talk directly and clearly to each other. This design helps agents ask questions, challenge each other, and give guidance more effectively. Tests showed BusMA works better than existing communication setups across various reasoning tasks.
What this means in practice
- •For software developers: Build complex AI systems that need multiple autonomous agents to coordinate and reason together more effectively using shared-channel communication.
- •For ai product teams: Create AI products requiring robust multi-agent communication, such as virtual assistants that combine separate reasoning tools seamlessly.$Commercial implications: Enables development of integrated assistant platforms combining multiple intelligent agents for enhanced problem solving.
Authors
Yanwen Peng, Delvin Ce Zhang, Xi Wang, Nikolaos Aletras
Abstract
Multi-Agent (MA) systems are effective at solving complex tasks that demand planning, tool use, and the synthesis of evidence from multiple sources. Existing systems typically adopt Hierarchical Manager-Worker (HMW) or Router-based Message Passing (RMP) structures as their communication protocol. However, these designs restrict agent autonomy: Worker agents cannot directly consult specific "peers", and misrouted messages can propagate errors. Inspired by bus architectures in computer systems, we propose BusMA, a communication framework that allows any agent to address other agents through a shared channel, i.e., the Bus. It consists of agent registration, message routing, and shared memory management components. Worker agents, each equipped with tools, have their own local memory and can reason, act (tool usage), and communicate by posting shared messages with specific intents. We introduce four intents: discussion, challenge, guidance, and request for explanation, which support fine-grained communication among agents. A Chair agent monitors the shared memory to coordinate interactions and facilitate convergence among Workers. To evaluate the effectiveness of BusMA, we conduct extensive experiments with two frontier LLMs across 13 tasks spanning visual reasoning, mathematical reasoning, and knowledge retrieval demonstrate that BusMA consistently outperforms state-of-the-art HMW and RMP methods.