The increasing consumption of processed and ultra-processed foods poses significant public health challenges due to their association with non-communicable diseases. Although nutritional labeling aims to guide consumers, information is often difficult to interpret and underutilized in decision-making. This work proposes a conceptual framework that integrates optical character recognition (OCR), large language models (LLMs), knowledge graphs (KGs), and retrieval-augmented generation (RAG) to automate the extraction, structuring, and retrieval of nutrition information from product labels while providing conversational guidance through a chatbot interface. The framework is instantiated in a system tailored to the Ecuadorian context, including a mobile application for product label capture, a web application for data management and verification, and a conversational assistant for natural language queries. Combining automation with structured representation and conversational guidance improves the accuracy, interpretability, and accessibility of nutritional data. Preliminary evaluations suggest that the framework supports efficient data management and improves the user’s understanding of nutritional information, contributing to intelligent and transparent health-oriented systems.