SpeakOne: Multilingual AI Chatbot for Student Support

Abstract

Higher education organizations face a huge number of repetitive inquiries from students regarding admissions, fee payments, exams, scholarships, placements, and other campus-related aspects. Inquiries are made in various languages, resulting in communication difficulties and increased work pressure on the administrative staff. This paper introduces the concept of SpeakOne, which is an AI chatbot for higher education organizations to answer student inquiries using a single conversational pipeline automatically. The designed system has features including automatic language detection, natural language processing, intent recognition, institutional knowledge base lookup, and an external generative-AI fallback module. Whenever there is a suitable institutional response to an inquiry, it is fetched by the system; otherwise, it is forwarded to an external AI language model. The recognized language is used throughout the processing pipeline to generate appropriate answers in the corresponding language. The system design has included conversation logging, user feedback, analytics, and an administrative dashboard that monitors unanswered questions and manages the knowledge base. The operational system dashboard indicates 15,234 queries, five languages, 95% success indicator for the system as a whole, and 45 unanswered queries; but these numbers refer to the operational system data and not machine learning benchmark-based accuracy levels.