Ask-before-Plan: Proactive Language Agents for Real-World Planning

TL;DR

The paper tackles the problem of proactive planning with language agents that must interpret ambiguous user instructions and react to environmental feedback. It introduces the Clarification-Execution-Planning (CEP) framework, composed of dedicated clarification, execution, and planning agents, augmented by trajectory tuning and memory recollection to improve reliability in real-world planning scenarios. A new dataset, Ask-before-Plan, extends TravelPlanner with uncertain instructions and proactive dialogues, enabling comprehensive evaluation across clarification, tool use, and planning. Extensive experiments show CEP improves clarification quality and robustness in tool learning and planning, while also revealing challenges in end-to-end planning and the benefits of environment-informed Clarification in guiding subsequent decisions.

Abstract

The evolution of large language models (LLMs) has enhanced the planning capabilities of language agents in diverse real-world scenarios. Despite these advancements, the potential of LLM-powered agents to comprehend ambiguous user instructions for reasoning and decision-making is still under exploration. In this work, we introduce a new task, Proactive Agent Planning, which requires language agents to predict clarification needs based on user-agent conversation and agent-environment interaction, invoke external tools to collect valid information, and generate a plan to fulfill the user's demands. To study this practical problem, we establish a new benchmark dataset, Ask-before-Plan. To tackle the deficiency of LLMs in proactive planning, we propose a novel multi-agent framework, Clarification-Execution-Planning (\texttt{CEP}), which consists of three agents specialized in clarification, execution, and planning. We introduce the trajectory tuning scheme for the clarification agent and static execution agent, as well as the memory recollection mechanism for the dynamic execution agent. Extensive evaluations and comprehensive analyses conducted on the Ask-before-Plan dataset validate the effectiveness of our proposed framework.

Figures (7)

• Figure 1: Illustrations of different problems.
• Figure 2: Data annotation process.
• Figure 3: Overview of the CEP framework.
• Figure 4: Analysis of clarification recall.
• Figure 5: Dependency graph for indefinite details.