The growing adoption of Generative AI is transforming how organizations design conversational systems. Large Language Models (LLMs) make it possible to build chatbots that understand natural language and handle a wider range of user requests compared to traditional rule-based systems. However, their use in enterprise contexts also introduces important challenges, especially in terms of data governance and reliability. This thesis explores the design and implementation of a domain-specific informational chatbot based on a Retrieval-Augmented Generation (RAG) architecture. The proposed system is designed as a first-line assistant that automatically answers user questions only by using information retrieved from an authorized knowledge base.