Do We Always Need Query-Level Workflows? Rethinking Agentic Workflow Generation for Multi-Agent Systems
Zixu Wang, Bingbing Xu, Yige Yuan, Huawei Shen, Xueqi Cheng · 2026
Read Do We Always Need Query-Level Workflows? Rethinking Agentic Workflow Generation for Multi-Agent Systems by Zixu Wang, Bingbing Xu, Yige Yuan, Huawei Shen, Xueqi Cheng (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-11147.txt.
Description
Zixu Wang, Bingbing Xu, Yige Yuan, Huawei Shen, and Xueqi Cheng rethink agentic workflow generation for multi-agent systems. Such systems solve tasks by coordinating agents through workflows. Workflows can be generated per task or per query. Query-level generation tailors each run, but its costs and benefits against task-level design were unclear. The authors analyze both, with theory and experiments.
They find that query-level generation is not always needed. A small set of the best task-level workflows already covers as many or more queries. That finding can cut the cost of running agent systems sharply. It tells architects to spend on a few strong workflows rather than bespoke ones for every request.
Catalog record
| CREATOR | Zixu Wang, Bingbing Xu, Yige Yuan, Huawei Shen, Xueqi Cheng |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 64 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-11147