Towards Multi-Turn Dialog Systems for Industrial Asset Operations and Maintenance

2026-05-24

AI Summary

This work addresses the challenges of industrial asset operation and maintenance question answering, which involves multi-turn iterative interactions and heavy reliance on external tool invocations, rendering traditional single-agent architectures ineffective at preserving cross-turn context and reusing intermediate results. To overcome these limitations, the authors propose a supervisor-expert multi-agent collaborative framework that enables efficient context-aware dialogue management through structured reuse of intermediate artifacts, dynamic task replanning, and parallelized tool execution. Experimental results demonstrate that the proposed approach improves task planning effectiveness by 54.5% and increases task completion rate by 37.8%. Furthermore, the proportion of time spent on tool invocation decreases from 47.3% to 26.3%, and response latency from the second to fifth turns is reduced to 4.2 times faster than that of the first turn.

Abstract

Industrial asset operations and maintenance question answering is inherently multi-turn, iterative, and highly dependent on external tool invocation. However, the conventional plan-execute single-agent architecture exhibits clear limitations in maintaining cross-turn context, and reusing intermediate results. In this paper, we present a multi-turn dialog system designed for industrial scenarios based on a supervisor-specialist multi-agent architecture. To alleviate tool invocation bottlenecks, the system incorporates structured artifact reuse, dynamic replanning, and parallel tool execution. Evaluation results show that our system achieves better response quality compared with the baseline, with planning effectiveness increasing by 54.5% and task completion improving by 37.8%. System profiling further shows that cross-turn artifact reuse effectively reduces redundant tool invocation, decreasing the tool-time share from 47.3% to 26.3% and making turns 2-5 approximately 4.2x faster than the first turn.

Problem

• multi-turn dialog
• industrial asset operations
• tool invocation
• context maintenance
• intermediate result reuse

Innovation

• multi-agent architecture
• structured artifact reuse
• dynamic replanning
• parallel tool execution
• multi-turn dialog system