WorkArena: How Capable Are Web Agents at Solving Common Knowledge Work Tasks?

arXiv:2403.07718v1 Announce Type: new
Abstract: We study the use of large language model-based agents for interacting with software via web browsers. Unlike prior work, we focus on measuring the agents' ability to perform tasks that span the typical daily work of knowledge workers utilizing enterprise software systems. To this end, we propose WorkArena, a remote-hosted benchmark of 29 tasks based on the widely-used ServiceNow platform. We also introduce BrowserGym, an environment for the design and evaluation of such agents, offering …

abstract agents arxiv browsers cs.ai cs.lg daily enterprise enterprise software focus knowledge knowledge workers language language model large language large language model measuring prior software study systems tasks type via web web browsers work workers