Zixu Wang, Bingbing Xu, Yige Yuan, Huawei Shen, Xueqi Cheng
Query-level workflow generation in Multi-Agent Systems may not be necessary, as task-level workflows can be equally effective and more efficient using the proposed SCALE framework.
In systems where multiple AI agents work together to solve tasks, it's common to create detailed plans, or workflows, for each query they handle. However, this study suggests that creating a few high-quality workflows for the overall task can be just as effective and much more efficient. The authors introduce a new method called SCALE, which uses a smarter way to predict and evaluate task-level workflows without exhaustive testing. This approach maintains nearly the same level of performance while significantly reducing the computational resources needed.
