Heart disease is a leading cause of global mortality, making early and reliable risk assessment critical for prevention and clinical decision support. Traditional machine learning models often provide high predictive performance but lack transparency in explaining their decisions. Therefore, there is a growing need for intelligent systems that combine predictive accuracy with explainable outputs suitable for real-world healthcare applications. This study presents a computer-based system that helps estimate a person’s risk of heart disease using common health information. The system also explains the results in simple, easy-to-understand language by using relevant medical knowledge. The framework combines a trained machine learning model for heart disease risk prediction with a Large Language Model Meta AI (LLaMA)-based large language model enhanced using an agentic Retrieval-Augmented Generation (RAG) mechanism. The RAG component retrieves relevant clinical context to ground explanations, improving clarity, consistency, and safety. Experimental evaluation shows that the system provides accurate risk predictions, clear and contextual explanations, and a smooth user experience. Agentic RAG improves explanation relevance and grounding, while LangGraph enhances robustness, fault tolerance, and execution traceability compared to linear pipelines. Experimental evaluation demonstrates strong predictive performance of the proposed model, achieving an accuracy of 88%, precision of 85%, recall of 82%, and an F1-score of 83.5%, while delivering clear, knowledge-grounded explanations through the integration of Agentic RAG and LangGraph orchestration. The results demonstrate the effectiveness of combining machine learning, agentic RAG-enabled large language models develop trustworthy and deployable healthcare decision-support systems for cardiovascular risk assessment.