Work agents unify AI models and systems for reliable digital tasks
Work Agents: Systems, Capabilities, and Optimization
Machine Learning
Summary
As AI models get better at doing digital tasks, it’s not clear how they complete complex jobs across many tools reliably. The authors look at work agents as systems combining AI models and software tools that keep track of ongoing tasks and produce checking outcomes. They classify different kinds of digital work and identify the key abilities agents need to do their jobs well. The paper also discusses how to improve both the AI and software parts together and how to measure success by what the work achieves. This helps guide building more trustworthy AI helpers for complex, practical tasks.
What this means in practice
- •For enterprise software teams: Develop AI-enhanced tools that reliably perform complex workflows by integrating foundation models with software execution systems.
- •For workflow automation engineers: Create and optimize agents that manage ongoing tasks across diverse applications with verifiable outcomes.
A position paper. It proposes an approach and reports no results.
Authors
Zishan Xu, Enjun Du, Mingqi Gao, Cheng Yang, Junhao Shen, Zirong Chen, Zhiyuan Yao, Yifu Guo, Yuquan Lu, Jinyang Huang, Zhengxi Lu, Siyi Liu, Ziyu Zheng, Yongqi Zhang, Difan Zou, Weiwen Liu
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.