Cocoa: Co-Planning and Co-Execution with AI Agents

2024-12-14

AI Summary

To address weak controllability and insufficient structuring in complex, multi-step document tasks, this paper proposes the “Interactive Planning” paradigm, enabling dynamic task-level human–AI collaboration—including co-planning and co-execution. Methodologically, it integrates computational notebook principles, an interactive document editing framework, and a lightweight AI agent scheduling mechanism to support visual plan construction and step-level real-time collaboration. A user study demonstrates that, compared to conventional chat-based interaction, Interactive Planning significantly improves AI steerability (p < 0.01) without compromising usability. Empirically, it reveals complementary applicability boundaries: Interactive Planning excels for structured, constraint-heavy tasks, whereas chat interfaces remain preferable for open-ended exploratory tasks. The core contribution is the first task-level human–AI co-design pattern, overcoming inherent limitations of conversational interfaces in structural expressivity and user control.

Technology Category

• Humans and AI: Planning and Decision Support for Human-Machine Teams
• Planning, Routing, and Scheduling: Planning with Language Models
• Multiagent Systems: Multiagent Planning

Application Category

• Search and Retrieval-Augmented AI: Assisted, interactive, and conversational search
• Economics, Online Markets and Human Computation: Research challenges in human and human-AI computation
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems

Abstract

We present Cocoa, a system that implements a novel interaction design pattern -- interactive plans -- for users to collaborate with an AI agent on complex, multi-step tasks in a document editor. Cocoa harmonizes human and AI efforts and enables flexible delegation of agency through two actions: Co-planning (where users collaboratively compose a plan of action with the agent) and Co-execution (where users collaboratively execute plan steps with the agent). Using scientific research as a sample domain, we motivate the design of Cocoa through a formative study with 9 researchers while also drawing inspiration from the design of computational notebooks. We evaluate Cocoa through a user study with 16 researchers and find that when compared to a strong chat baseline, Cocoa improved agent steerability without sacrificing ease of use. A deeper investigation of the general utility of both systems uncovered insights into usage contexts where interactive plans may be more appropriate than chat, and vice versa. Our work surfaces numerous practical implications and paves new paths for interactive interfaces that foster more effective collaboration between humans and agentic AI systems.

Problem

• Human-AI Collaboration
• Document Processing
• Complex Task Planning

Innovation

• Cocoa System
• Human-AI Collaboration
• Enhanced Efficiency