Large Language Models-Based Robots for Enterprise Financial Statement Auditing

This article aims to enhance the digital and intelligent capabilities of enterprise financial statement audits by proposing and constructing optimization strategies and frameworks for audit robots based on large language models. It utilizes large language models to design the audit data robot, audit workpaper robot, and audit analysis robot for human-computer interaction, thereby lowering the technical threshold for auditors and improving their understanding of personalized audit requirements. The audit data robot intelligently extracts data through natural language commands, the audit workpaper robot automates the preparation of workpapers through intent recognition and tool library selection, and the audit analysis robot enhances decision-making levels using visualization and intelligent analysis techniques. The audit robots constructed based on large language models will effectively address existing issues in current audits and improve audit efficiency and intelligence levels.