IntentCUA: Learning Intent-level Representations for Skill Abstraction and Multi-Agent Planning in Computer-Use Agents

Researchers introduced IntentCUA, a multi-agent framework for computer automation that achieved 74.83% task success rate through intent-aligned planning and memory systems. The system uses coordinated agents (Planner, Plan-Optimizer, and Critic) to reduce error accumulation and improve efficiency in long-horizon desktop automation tasks.

• →IntentCUA framework achieved 74.83% task success rate with 0.91 step efficiency ratio, outperforming existing RL-based and trajectory methods
• →The system uses three coordinated agents sharing memory to abstract interaction traces into reusable skills and intent representations
• →Multi-view intent abstraction and shared plan memory jointly improve execution stability in desktop automation
• →The approach reduces redundant re-planning and mitigates error propagation across desktop applications
• →Research demonstrates that system-level intent abstraction is key to reliable automation in dynamic environments