Design and Implementation of a Knowledge-Based AI Support Agent for Real-Time Calls

Read full text Summarise with VIRAT
Authors: S.V. Nagaraj, M.Birunthadevi, R.Kavitha, Sandhiya S, J.Subahsri
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper presents an AI-Powered College Inquiry Agent designed to handle real-time inbound calls from prospective students and stakeholders in an automated and efficient manner. Incoming calls from customers are first received by the AI agent, which orchestrates the end-to-end interaction workflow. Audio data is sourced and streamed using the integration of Twilio and LiveKit, enabling reliable and low-latency voice communication. The incoming speech is processed through a speech-to-text module to generate realtime transcriptions, which are then passed to a response management module.The response management module leverages a centralized knowledge base combined with AI logic to generate accurate, consistent, and context-aware responses to college-related inquiries such as admissions, courses, and general information. In cases where the agent is unable to confidently answer a query, the system automatically forwards the call details and transcribed conversation in text format to the respective department head for further review. If the head identifies the inquiry as relevant or high-priority, they may directly contact the customer to continue the interaction. The generated responses are converted into natural-sounding speech using a text-tospeech engine and delivered to the customer in real time. This approach ensures continuity of communication while reducing dependency on live human intervention. The proposed system enhances inquiry handling efficiency and improves decision-making in academic communication workflows.

Fields of Study & Subjects

AI in Service Interactions Speech and dialogue systems Artificial Intelligence Applications

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
1 January 2026
Language
EN