This master's thesis explores the development and implementation of an intelligent travel assistant designed to enhance user travel experiences through the integration of advanced natural language processing techniques. The core focus of this research is the utilization of Retrieval-Augmented Generation (RAG) technique combined with large language models (LLMs) to deliver accurate, contextually relevant information and recommendations. The thesis begins by discussing key concepts crucial to understanding the technology behind the travel assistant, including large language models, the hallucination problem, RAG, vector embeddings, and vector databases. It provides a overview of existing applications that utilize RAG extensions currently. In the technology section, the thesis elaborates the components and tools used in the proposed solution, such as TravelTalk, WikiTravel, LangChain, PineCone, and FastAPI. The methodology outlines the step-by-step process, starting from web scraping and data preparation, embedding and upserting (Pinecone terminology used for simultaneous inserting and updating) vectors into a vector database, to connect the OpenAI API to the vector database for retrieval, and finally, developing and integrating an API with a frontend application. Results from the implementation demonstrate the effectiveness of the personalized travel assistant in providing users with tailored travel recommendations and information. The thesis concludes with an analysis of the project's outcomes, potential improvements, and future directions for enhancing the assistant's capabilities. This thesis contributes to the field of AIdriven travel assistance by showcasing the practical application of LLMs and RAG, offering valuable insights for future developments in personalized AI solutions.