Overview
Research area: ML Authors: Zishan Xu, Enjun Du, Mingqi Gao, Cheng Yang, Junhao Shen et al. Published: 2026-10-26 arXiv: 2610.35845
Abstract (translated)
As foundation models evolve from information generation toward digital work execution, agents are increasingly capable of completing complex tasks across tools, applications, and workflows. Yet existing research lacks a unified perspective on how agents reliably accomplish delegated work. This paper proposes a work-centric, system-level view: a work agent is a system combining a foundation model with a runtime harness that executes tasks in a stateful workspace and delivers verifiable outcomes.
The authors introduce two complementary dimensions—work persistence and work openness—to characterize working conditions, and classify digital work into five forms according to task scope and organization.
On top of this framework, the paper:
- Identifies five core capability families required for reliable work execution
- Examines optimization approaches: model optimization, harness optimization, and their joint optimization
- Analyzes existing evaluation methods and metrics from the perspective of work outcomes
- Discusses future directions: sustained work delegation, adaptation to open environments, and the interaction between work execution and capability improvement
Original Abstract
> As foundation models evolve from information generation to digital work execution, agents are increasingly capable of completing complex tasks across tools, applications, and workflows. However, existing research lacks a unified perspective on how agents reliably accomplish delegated work. To address this gap, we present a work-centric, system-level perspective that views work agents as systems combining foundation models and runtime harnesses to execute tasks in stateful workspaces and deliver verifiable outcomes. We introduce two complementary dimensions, work persistence and work openness, to characterize working conditions, and classify digital work into five forms according to task scope and organization. Building on this framework, we identify five core capability families required for reliable work execution and examine approaches to model optimization, harness optimization, and their joint optimization. We further analyze existing evaluation methods and metrics from the perspective of work outcomes. Finally, we discuss future research directions, including sustained work delegation, adaptation to open environments, and the interaction between work execution and capability improvement. This paper provides a unified framework for understanding, optimizing, and evaluating work agents, supporting the development of systems capable of delivering reliable, verifiable, and practically valuable work outcomes.
--- *Auto-collected on 2026-10-28*