The integration of artificial intelligence in tourism has revolutionized visitor engagement by offering interactive, personalized, and real-time information services. This paper presents the development of an intelligent tourist chatbot system tailored for Mirandela, a culturally rich destination in northern Portugal. The system combines advanced web scraping techniques and natural language processing to deliver relevant and context-aware recommendations to tourists. A specialized web crawler was developed to extract up-to-date content from the official VisitMirandela website, which is then processed and stored using a hybrid architecture integrating markdown and vector databases. User queries are handled through a structured response pipeline utilizing OpenAI’s GPT-4o Mini model to generate semantically accurate and sourced answers. The system was evaluated through user-based question testing, achieving high accuracy in most scenarios while revealing opportunities for enhancement, particularly in providing named recommendations. The results highlight the potential of AI chatbots to enrich the visitor experience, improve access to local information, and reduce the dependence on multiple digital platforms. Furthermore, the paper addresses key challenges such as preserving human interaction, ensuring data reliability, and maintaining cultural authenticity. The proposed architecture demonstrates a scalable and effective approach for implementing AI-driven conversational systems in local tourism strategies, contributing to the digital transformation of the sector. Future work includes expanding the knowledge base, incorporating multimodal interactions, and enhancing emotional intelligence in chatbot responses.