# ── CITATION.cff ───────────────────────────────────────────────
cff-version: 1.2.0
title: "OWL: Optimized Workforce Learning for General Multi-Agent Assistance in Real-World Task Automation"
message: "If you use OWL or find it helpful, please cite this paper."
abstract: >
  Large Language Model (LLM)-based multi-agent systems show promise for
  automating real-world tasks but struggle to transfer across domains
  due to their domain-specific nature. We introduce Workforce, a
  hierarchical multi-agent framework that decouples strategic planning
  from specialized execution via a domain-agnostic Planner, a
  Coordinator, and domain-specific Worker agents. Optimized Workforce
  Learning (OWL) further improves cross-domain generalization through
  reinforcement learning from real-world feedback. Experiments on the
  GAIA benchmark show state-of-the-art open-source performance
  (69.70 %), surpassing commercial systems like Deep Research and
  approaching GPT-4o. By enabling scalable generalization and modular
  domain transfer, OWL lays a foundation for the next generation of
  general-purpose AI assistants.
version: "v2"          
doi: 10.48550/arXiv.2505.23885
url: https://arxiv.org/pdf/2505.23885
date-released: 2025-06-11

authors:
  - family-names: Hu
    given-names: Mengkang
  - family-names: Zhou
    given-names: Yuhang
  - family-names: Fan
    given-names: Wendong
  - family-names: Nie
    given-names: Yuzhou
  - family-names: Xia
    given-names: Bowei
  - family-names: Sun
    given-names: Tao
  - family-names: Ye
    given-names: Ziyu
  - family-names: Jin
    given-names: Zhaoxuan
  - family-names: Li
    given-names: Yingru
  - family-names: Chen
    given-names: Qiguang
  - family-names: Zhang
    given-names: Zeyu
  - family-names: Wang
    given-names: Yifeng
  - family-names: Ye
    given-names: Qianshuo
  - family-names: Ghanem
    given-names: Bernard
  - family-names: Luo
    given-names: Ping
  - family-names: Li
    given-names: Guohao
# ───────────────────────────────────────────────────────────────