An agentic RAG approach for knowledge management in enterprise consulting contexts

Efficient access to corporate knowledge is a key challenge for modern organizations, especially in technical environments such as consulting and software development. Information is often distributed across heterogeneous tools and repositories, forcing employees to spend significant time searching for internal documentation and interrupting their workflow. Traditional search systems therefore present limitations in terms of efficiency and usability. This thesis presents the design, development, and evaluation of an intelligent assistant aimed at improving access to internal knowledge within Blue Reply, an IT consulting company operating in the Italian enterprise sector. The proposed system adopts a modular architecture that integrates document retrieval with generative artificial intelligence, enabling responses to be grounded in verifiable and updatable sources. The system was evaluated using automatic metrics and user feedback to assess both retrieval effectiveness and response quality. The results show high levels of accuracy, consistency with source documents, and practical usefulness for users. Overall, the proposed approach demonstrates that a modular retrieval-augmented architecture can effectively support knowledge access in enterprise environments, reducing information search time and improving operational efficiency. A limitation of the study is that the evaluation relies partly on automatic metrics and a synthetic dataset, which may require further validation through larger-scale studies and expert assessment.