An AI-Enabled Dual-Database Chatbot for Academic Advising and Training Management in Distance Higher Education

This article presents the design, implementation, and initial evaluation of an artificial-intelligence-enabled chatbot for academic advising and training management in a university distance-learning context. The study addresses a practical management problem: distance learners often need timely, accurate, and personalized answers about regulations, curricula, credit registration, learning progress, assessment, and graduation conditions, while academic advisors face high volumes of repetitive inquiries during peak periods. The proposed solution combines Retrieval-Augmented Generation (RAG), a vector database for unstructured institutional documents, a relational database for structured student and training records, intent classification, prompt engineering, guardrails, and a web-based chat interface. The system transforms formal documents such as training regulations, advisor handbooks, curricula, and appendices into semantic embeddings, retrieves relevant passages at query time, and generates grounded answers with a lower risk of hallucination. For personalized advising, structured data are queried through a controlled relational layer. Initial evaluation shows approximately 85% accuracy for regulation-related questions, an average response time of about five seconds, effective storage of conversation history, and an estimated 60-70% reduction in repetitive manual advising workload. The contribution of the article is a practical dual-database architecture and implementation framework that can support digital transformation in training management, especially for open and distance education institutions in contexts where policy documents and student records must be handled together.