This undergraduate thesis presents the objective of developing an intelligent assistant aimed at analysis and decision support in the financial market, a sector characterized by a massive volume of data. The method employed to solve this problem is the implementation of a RAG, Retrieval-Augmented Generation architecture. This architecture utilizes a vector database to allow a Large Language Model to access real-time information, overcoming its static knowledge. The vector database is continuously fed by multiple data sources, integrating public data (recent news via API, social media posts via web scraping) with the user s private data (uploaded PDF reports and the assets composing their portfolio). The results are demonstrated through a functional prototype that allows the user to not only manage their portfolio but also interact with the assistant. The system is capable of interpreting natural language queries and using software tools to retrieve relevant context (whether from the vector database or APIs) before formulating a response. It is concluded that the use of RAG architectures applied to finance is a viable and effective approach. The system demonstrated the ability to integrate disperse information and offer personalized analytical support, broadening access to information and enhancing the understanding of complex market scenarios.