Generative AI-Driven Rule Engines for IoT Platforms : A Design Science Research (DSR) Study

Abstract

The Internet of Things (IoT) has brought immense potential for automation and connectivity, but its complexity poses significant challenges, particularly for non-technical users. Managing an IoT system often requires technical expertise, especially when configuring rule engines, a critical component that automate tasks based on predefined conditions. For individuals without programming skills, setting up and managing these rule engines can be difficult, leading to a steep learning curve and limiting accessibility to IoT technology. This thesis focuses on addressing these challenges by integrating Generative AI, specifically ChatGPT, into an IoT platform through the development of a Smart Rule Engine tool. The tool simplifies the configuration process, enabling users to create and manage rule chains using natural language commands rather than complex workflows and coding. This approach empowers users with domain expertise but no or limited IoT knowledge to independently manage IoT systems. The research demonstrates that ChatGPT’s natural language processing capabilities can effectively translate user commands into machine-readable rule chains. However, the study also identifies challenges such as occasional inaccuracies in understanding user intent and generating incorrect outputs. Solutions such as prompt engineering and model fine-tuning are proposed to improve system performance. In addition to simplifying rule engine configuration, this work paves the way for simplifying device provisioning, dashboard creation, and ultimately controlling the entire IoT platform through the chat interface. The findings contribute to the fields of IoT, End-User Development (EUD), and Natural Language Interfaces (NLI), illustrating the potential of AI-driven automation to lower barriers for non-technical users in managing complex IoT environments.