RAG-LLM Chatbot for Halal Certification 4.0: Enhancing Indonesian Halal Regulatory Literacy and Information Access

The increasing complexity of halal certification processes in Indonesia, alongside the growing demand for accessible and trustworthy information, highlights the need for intelligent. This study presents the design, integration, and evaluation of a Retrieval-Augmented Generation (RAG)-based chatbot system powered by a Large Language Model (LLM) to support Halal Certification 4.0. The system architecture was developed using LangChain and OpenAI GPT-4, with a modular pipeline integrating official halal documents from SIHALAL, fatwas issued by the Indonesian Ulema Council (MUI), and national halal regulations. The chatbot allows users to interact in Bahasa Indonesia via web or telegram, enabling real-time halal consultation by retrieving and generating responses grounded in authoritative sources. The evaluation was conducted using 100 document-based queries covering certification data, religious rulings, and regulatory content. The chatbot achieved an overall accuracy of 87%, with the highest accuracy in SIHALAL-related queries (91%) and strong performance in regulation-based responses (88%). This research demonstrates how AI can be ethically and effectively deployed to improve public access to halal information, promote regulatory literacy among MSMEs, and enhance transparency in Islamic service delivery. The proposed system sets a practical foundation for future developments in AI-assisted halal governance in Indonesia.