Seoyoung Lee, Seobin Yoon, Seongbeen Lee, Yoojung Chun, Dayoung Park, Doyeon Kim, Joo Yong Sim
IntentCUA is a framework that improves computer-use agents' task success and efficiency by using intent-level representations and shared plan memory for skill abstraction and multi-agent planning.
IntentCUA is a new approach to making computer-use agents more effective and efficient when performing tasks on a computer. Traditional methods often struggle with long and complex tasks, leading to mistakes and inefficiency. IntentCUA addresses this by using a shared memory system where different agents work together to understand the user's intent and reuse skills, reducing the need to plan from scratch each time. This makes the agents better at completing tasks successfully and with fewer errors.
