StateAct: Program State, before Pixels, for Long-Horizon Computer-Use Agents

2026-07-24

AI Summary

This work addresses the limitations of conventional agents that rely heavily on screen pixel perception, which often leads to poor accuracy in understanding and manipulating underlying program states, resulting in low success rates and high costs for long-horizon tasks. To overcome this, the authors propose a program-state-centric multi-agent framework in which a primary agent directly operates on program states—such as the file system, application backends, and the DOM—through code, invoking a lightweight GUI sub-agent only when necessary for interface interactions. An independent verification mechanism is integrated to ensure correctness. By anchoring action, verification, and memory in program states rather than visual pixels, this approach establishes a state-driven reasoning paradigm that substantially reduces reliance on visual perception. Evaluated on OSWorld 2.0, it improves Claude Opus’s binary success rate from 20.6% to 26.9% and partial success rate from 54.8% to 61.6%, while reducing task cost by approximately 9×.

Abstract

Computer-use agents are usually improved by strengthening perception: better models for reading a screenshot and choosing where to click. Yet a screenshot is only a lossy rendering of the underlying program state, e.g., the files, application backends, and DOM that hold the task data. Different states can produce the same pixels, while code can inspect and modify that state directly. StateAct is a code-first, multi-agent harness built around this distinction. Its main agent works directly with program state by using code, while a dedicated GUI subagent handles screenshot-and-click interaction on the few subgoals that need it, just 28 of 108 tasks and 1.1% of main-agent steps. The same direct access to program state also supports verification: an independent finish gate double-checks the saved result for structural failures, e.g., output that is missing, unsaved, or written to the wrong path. To stay on track over hundreds of steps, the main agent hands subgoals to fresh subagents, keeping its own context focused. On OSWorld 2.0, StateAct lifts Claude Opus 4.8 from 20.6% to 26.9% on binary success, and from 54.8% to 61.6% on partial success, at ~ 9x lower cost per task than the same model driven by screenshots alone; a code-only variant with no GUI subagent reaches only 45.9% partial, below that screenshot-based baseline's 54.8%. In general, grounding action, verification, and memory in state, what we call state-grounding, shifts the main bottleneck from perception toward reasoning: failures depend more on what the agent thinks than on what it sees.

Problem

• program state
• computer-use agents
• long-horizon tasks
• perception bottleneck
• state-grounding

Innovation

• state-grounding
• program state
• multi-agent architecture
• computer-use agents
• verification