Tourism information services in Majalengka Regency face challenges regarding limited operational hours and repetitive administrative workloads. This study aims to design and implement a web-based tourism information system equipped with a Large Language Model (LLM) virtual assistant to automate 24/7 information delivery. The development utilized the Waterfall method, encompassing requirements analysis, design, implementation using Next.js and Vercel AI SDK, and testing. To mitigate the risk of AI hallucination, Context Grounding was applied using local JSON data containing official tourism details. Black-box testing results demonstrate that the system functions optimally, with the AI successfully answering factual queries accurately within two seconds, rejecting out-of-context topics, and improving overall operational efficiency
Copyrights © 2026