CoAct-1: Computer-using Agents with Coding as Actions

2025-08-05

AI Summary

Existing GUI-based autonomous agents suffer from inherent inefficiency and poor robustness in complex, long-horizon tasks. To address this, we propose a multi-agent system that synergistically integrates GUI interaction with programmatic execution—introducing “write-and-execute code” as a primitive action within the agent’s action space for the first time. A dynamic task dispatching mechanism, governed by a centralized coordinator, intelligently selects between GUI operations and script execution based on task context, enabling adaptive, complementary utilization of both modalities. The system incorporates vision-language understanding, Python/Bash script generation, and secure local execution. Evaluated on OSWorld, it achieves a new state-of-the-art success rate of 60.76%, reduces average task steps to 10.15—a 32.3% improvement over prior best—and significantly enhances automation depth, reliability, and generalization across diverse desktop environments.

Technology Category

• Multiagent Systems: Adversarial Agents
• Humans and AI: Planning and Decision Support for Human-Machine Teams
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Autonomous agents that operate computers via Graphical User Interfaces (GUIs) often struggle with efficiency and reliability on complex, long-horizon tasks. While augmenting these agents with planners can improve task decomposition, they remain constrained by the inherent limitations of performing all actions through GUI manipulation, leading to brittleness and inefficiency. In this work, we introduce a more robust and flexible paradigm: enabling agents to use coding as a enhanced action. We present CoAct-1, a novel multi-agent system that synergistically combines GUI-based control with direct programmatic execution. CoAct-1 features an Orchestrator that dynamically delegates subtasks to either a conventional GUI Operator or a specialized Programmer agent, which can write and execute Python or Bash scripts. This hybrid approach allows the agent to bypass inefficient GUI action sequences for tasks like file management and data processing, while still leveraging visual interaction when necessary. We evaluate our system on the challenging OSWorld benchmark, where CoAct-1 achieves a new state-of-the-art success rate of 60.76%, significantly outperforming prior methods. Furthermore, our approach dramatically improves efficiency, reducing the average number of steps required to complete a task to just 10.15, compared to 15 for leading GUI agents. Our results demonstrate that integrating coding as a core action provides a more powerful, efficient, and scalable path toward generalized computer automation.

Problem

• Improving efficiency and reliability of GUI-based autonomous agents
• Combining GUI control with programmatic execution for flexibility
• Reducing task completion steps through coding-enhanced actions

Innovation

• Combines GUI control with programmatic execution
• Dynamic delegation to GUI Operator or Programmer
• Uses Python/Bash scripts for efficient task handling