通过真实流量自动生成专家代理画像，实现动态任务路由优化。
FlyRoute: Self-Evolving Agent Profiling via Data Flywheel for Adaptive Task Routing
- 基于真实请求自动构建代理能力证据，形成数据飞轮。
- 仅用5个种子查询，路由准确率从72.57%提升至89.83%。
- 适合需要持续优化智能客服或开发支持系统的团队。
企业路由器将查询分配给专家代理，但部署的代理画像长期不变，而代理本身（提示、工具、模型）持续演进，开发者也极少更新描述或示例。我们提出FlyRoute，一种自演化画像框架：从真实流量中积累候选分派结果与成功配对，定期将证据提炼为能力描述，并注入到基于BM25检索的成功案例与大语言模型路由器中。为提升数据效率，FlyRoute引入目标探索策略，结合画像不确定性、BM25相关性与词汇新颖性，仅对可能匹配的查询优先探索未充分画像的代理，避免冗余收集。在自有企业开发者支持数据集上实验，使用每代理仅5个种子查询，路由准确率从72.57%提升至89.83%（+17.26pp），跨四个专业领域均表现稳定，证明画像检索可显著增强冷启动路由性能。
原文摘要 · Abstract (English)
Enterprise routers assign queries to expert agents, yet deployed profiles stay static while agents evolve (prompts, tools, models), and developers rarely keep descriptions or exemplars current. We present FlyRoute, a self-evolving profiling framework that grows capability evidence from real traffic: dispatch candidates, quality-gate successful pairs into each agent's success store, periodically distill evidence into learned capability descriptions, and inject those descriptions together with BM25-retrieved successes into an LLM router. To make this flywheel data-efficient, FlyRoute introduces a targeted exploration policy that combines profile uncertainty, BM25 relevance, and lexical novelty, prioritizing under-profiled agents only for plausible queries and avoiding redundant evidence collection. In experiments on our proprietary enterprise developer-support dataset of real routed queries, FlyRoute improves a same-backbone zero-shot LLM router from 72.57% to 78.04% with only five seed queries per agent, showing that profile retrieval already strengthens cold-start routing. After streaming 7,211 labeled training queries through the flywheel, accuracy rises to 89.83% (+17.26pp over zero-shot; +11.79pp over cold start), with consistent gains across four expert domains under standard routing accuracy on single-gold test queries.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。