TCAndon-Router: Adaptive Reasoning Router for Multi-Agent Collaboration
Jiuzhou Zhao, Chunrong Chen, Chenqi Qiao, Lebin Zheng, Minqi Han, Yanchi Liu Yongzhou Xu Xiaochuan Xu Min Zhang
2026-01-12
Summary
This paper introduces a new way to route questions to the best 'expert' within a system of many intelligent agents working together, called TCAndon-Router (TCAR). It's designed to be better at handling complex, real-world situations where different experts might have overlapping skills.
What's the problem?
Current methods for directing questions to the right expert agent are often too rigid. They struggle when new experts are added or when multiple experts could potentially answer the same question, leading to confusion and less accurate results. Existing systems either focus on speed and cost or on finding the most specialized expert, but they don't easily adapt to changes or handle overlapping expertise well.
What's the solution?
The researchers created TCAR, a router that can dynamically adjust as new experts join the system. It first thinks through the question in natural language to understand what's being asked, then identifies several agents who *could* handle it. These agents each provide an answer, and a special 'Refining Agent' combines those answers into a single, high-quality response. This approach avoids getting stuck on a single, potentially incorrect choice.
Why it matters?
This research is important because it makes multi-agent systems more practical for real-world applications, like customer service or data analysis. By allowing for easy updates and resolving conflicts between experts, TCAR improves accuracy and reliability, making these systems more trustworthy and useful. The code for TCAR has also been made publicly available to encourage further development in this area.
Abstract
Multi-Agent Systems(MAS) have become a powerful paradigm for building high performance intelligent applications. Within these systems, the router responsible for determining which expert agents should handle a given query plays a crucial role in overall performance. Existing routing strategies generally fall into two categories: performance routing, which balances latency and cost across models of different sizes, and task routing, which assigns queries to domain-specific experts to improve accuracy. In real-world enterprise applications, task routing is more suitable; however, most existing approaches rely on static single-label decisions, which introduce two major limitations: (i) difficulty in seamlessly integrating new agents as business domains expand, and (ii) routing conflicts caused by overlapping agent capabilities, ultimately degrading accuracy and robustness.To address these challenges, we propose TCAndon-Router(TCAR): an adaptive reasoning router for multi-agent collaboration. Unlike traditional routers, TCAR supports dynamic agent onboarding and first generates a natural-language reasoning chain before predicting a set of candidate agents capable of handling the query. In addition, we design a collaborative execution pipeline in which selected agents independently produce responses, which are then aggregated and refined into a single high-quality response by a dedicated Refining Agent.Experiments on public datasets and real enterprise data demonstrate that TCAR significantly improves routing accuracy, reduces routing conflicts, and remains robust in ambiguous scenarios. We have released TCAR at https://huggingface.co/tencent/TCAndon-Router to support future research on explainable and collaborative multi-agent routing.