LLMs as AI Middleware: Unifying Disparate Systems in Manufacturing IT Landscapes

Abstract

This research investigates the potential of Large Language Models (LLMs) as AI middleware to unify disparate systems within manufacturing IT landscapes. Traditional manufacturing enterprises often contend with siloed data, legacy systems, and heterogeneous interfaces that impede seamless integration and automation. By leveraging LLMs’ capabilities in natural language understanding, contextual reasoning, and semantic interoperability, organizations can facilitate intelligent translation, integration, and orchestration across core systems—including Human Resources (HR), Payroll, Enterprise Resource Planning (ERP), Sales Order Management (SOM), Retail Management Systems (RMS), and supply chain platforms. The study introduces a conceptual framework positioning LLMs as cognitive intermediaries, significantly reducing integration complexity and enhancing cross-system data flow. Through real-world manufacturing scenarios, the framework demonstrates improved agility, minimized manual configuration, and more intuitive human-system interaction across the manufacturing digital thread.