Khon Kaen University Integrated Generative AI Services System

Abstract

This research aims to develop an “Integrated Generative AI Service System” to serve as a centralized platform for Khon Kaen University (KKU) personnel and students, with the primary objectives of enhancing resource management efficiency, controlling AI utilization expenditures, and establishing a standardized framework for integration with the university's core information systems. The system was developed using a microservices architecture, dividing functionalities into independent services. The User Interface supports access via Web Browsers and Mobile Applications, while the Web Service layer comprises an API Server acting as the central hub for processing commands and managing business logic, interfacing with a diverse database infrastructure including MariaDB (Primary Database), Redis (Cache/Session Data), and Object Storage (File Server). At the core of the system is the Large Language Model (LLM) Integration component, which functions as a centralized gateway managing and interacting with various LLMs. It employs the OpenRouter LLMs Gateway as an intermediary to select and route requests to the most appropriate LLM provider (e.g., OpenAI, Google, Anthropic) to optimize for cost and processing speed. Furthermore, the system integrates authentication through KKU Single Sign On (SSO) and implements rigorous security measures, including HTTPS/TLS encryption, XSS/CSRF protection, Rate Limiting, and API Key management, to ensure secure and effective AI utilization. User satisfaction evaluation showed that user satisfaction with the Integrated Generative AI Service System at Khon Kaen University was at a high level (with an overall mean score of 4.18 and S.D. of 0.83).