Automating Complex Document Workflows via Stepwise and Rollback-Enabled Operation Orchestration

2025-12-03

AI Summary

This paper addresses the challenges of step-level control and irreversible errors in automating complex document workflows. We propose Stepwise-and-Rollback, a novel execution framework featuring instruction-guided action planning, intent-filtered API candidate generation, fine-grained document state tracking, and dual-level rollback—operating at both parameter and API granularity—to enable progressive multi-step orchestration and dynamic error correction. We introduce a new document-processing benchmark comprising 250 real-world dialogues, the first to support continuous alignment between user intent and document state over long-horizon tasks. Experiments show our method achieves 90% instruction-level and 62% session-level task completion rates—improving over baselines by 40% and 76%, respectively—while demonstrating robustness across diverse LLMs and task difficulty levels.

Technology Category

• Natural Language Processing: Safety and Robustness
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Workflow automation promises substantial productivity gains in everyday document-related tasks. While prior agentic systems can execute isolated instructions, they struggle with automating multi-step, session-level workflows due to limited control over the operational process. To this end, we introduce AutoDW, a novel execution framework that enables stepwise, rollback-enabled operation orchestration. AutoDW incrementally plans API actions conditioned on user instructions, intent-filtered API candidates, and the evolving states of the document. It further employs robust rollback mechanisms at both the argument and API levels, enabling dynamic correction and fault tolerance. These designs together ensure that the execution trajectory of AutoDW remains aligned with user intent and document context across long-horizon workflows. To assess its effectiveness, we construct a comprehensive benchmark of 250 sessions and 1,708 human-annotated instructions, reflecting realistic document processing scenarios with interdependent instructions. AutoDW achieves 90% and 62% completion rates on instruction- and session-level tasks, respectively, outperforming strong baselines by 40% and 76%. Moreover, AutoDW also remains robust for the decision of backbone LLMs and on tasks with varying difficulty. Code and data will be open-sourced. Code: https://github.com/YJett/AutoDW

Problem

• Automates multi-step document workflows with stepwise orchestration
• Enables dynamic correction via rollback mechanisms for fault tolerance
• Improves completion rates in realistic document processing scenarios

Innovation

• Stepwise API action planning with document state
• Robust rollback mechanisms at argument and API levels
• Dynamic correction and fault tolerance in workflows