evaluation-environment · Documentary review
WorkArena
Evaluates web agents on enterprise knowledge-work tasks.
OriginServiceNow Research and the WorkArena paper authorsTopicsAutonomy & agentsStatusReviewed
Can support
Performance on WorkArena's selected ServiceNow task distribution under the specified instance, agent interface, and budget.
Cannot support by itself
General enterprise productivity, organizational integration, safe access control, user acceptance, or performance across other business software.
Decision use
Best used for
Testing browser agents on reproducible enterprise workflows and compositional planning tasks.
Not enough for
Procurement claims that an agent can safely automate enterprise work without organization-specific pilots and controls.