This paper presents the design and implementation of a modular conversational agent architecture that enables natural language access to blockchain-based traceability data in the agro-food industry. The research addresses a critical challenge: end-users and consumers require access to traceability data but cannot feasibly receive training on complex technical systems. We propose an LLM-based chatbot that allows stakeholders to query supply chain information using natural language interactions. The architecture integrates two complementary execution backends - LangGraph for structured reasoning with state management, and Gemini Live API for real-time multilingual voice interactions - demonstrating that conversational interfaces can effectively bridge the gap between complex blockchain systems and non-technical users. Through the Model Context Protocol, the system connects heterogeneous data sources including documents, APIs, and blockchain event logs, while role-based access controls ensure appropriate information access across different stakeholder types. A proof-of-concept implementation validates the technical feasibility of this approach, demonstrating that end-users can access verifiable supply chain data through natural dialogue without prior training in blockchain technology or traceability protocols. This work establishes an architectural foundation for making decentralized traceability systems operationally accessible, eliminating the user training barrier that has hindered consumer-facing blockchain applications in agri-food supply chains.
Fernandes et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: