Enhancing Warehouse Operations with Voice-Activated SQL Streamlining Using LLM Technology

Enhancing Warehouse Operations with Voice-Activated SQL Streamlining Using LLM Technology

연구 분야: Databases

학회: International Advanced Computing Conference

Inventory management of warehouses has traditionally been a tedious task. This paper proposes a customized inventory management software solution that leverages Large Language Models (LLMs) to be operated through natural language inputs, accepting both text and voice commands. This eliminates the need for dedicated database expertise, helping companies organize their databases in real-time upon the arrival or dispatch of goods, thus saving hours of manual work. The system ensures user roles and restrictions to prevent unauthorized access and provides a rollback mechanism for unintended database alterations. Our solution enhances inventory management with advanced voice activation powered by LLMs, robust security protocols, and real-time database updates, as demonstrated through various performance metrics and interface snapshots. With the integration of LLMs, accuracy improved from 62% to 85%, and the error rate decreased from 10% to 1%. While response time increased due to additional processing, the overall efficiency and security of the database management were significantly enhanced, providing a comprehensive and secure solution for managing inventories of all sizes.