Processing healthcare data presents significant challenges for clinicians, particularly in chronic disease management such as diabetes care. Traditional data aggregation and summarization methods often struggle to integrate the diverse data sources of health information, ranging from structured electronic health records to unstructured clinical notes. In response to this challenge, the paper proposes a reference framework that leverages generative Artificial Intelligence (AI) techniques and agentic AI architectures for enhancing health data management and decision support. By employing Large Language Models, the system is able to synthesize insights from collected data also deriving contextualized summaries and personalized recommendations useful for healthcare professionals. The paper outlines the architectural components of the framework and discusses key implementation strategies. Finally, a basic prototype has been developed to demonstrate the feasibility and usability of the proposed architecture in the context of telediabetology, supporting the management of diabetes patients during periodic clinical visits.