UNIRI robot background image

UNIRI Sports Center
AI Assistant

AI assistant for the UNIRI Sports Center a modern, automated and intelligent platform that helps students, coaches and staff access schedules, facilities and smart recommendations in seconds. Best of all, it is self-hosted, which guarantees complete ownership over data privacy and integrity and helps minimize inaccuracy and hallucinations.

Overview

Technology stack illustration

UNIRI Center for Sports AI Assistant is a self-hosted RAG system for the UNIRI Center for Sports that helps students, coaches and staff instantly find the information they need: training schedules, rules and regulations for each sport, facility availability, sports events and internal communications.

The platform centralizes data, automates routine questions and delivers clear, human-friendly answers through an intuitive, responsive interface.

Challenge

Technology stack illustration

Previously, key information was scattered across PDFs, website pages, spreadsheets, Word documents, and physical notice boards, making it hard for students and staff to get quick answers to everyday questions. The goal was to build an AI assistant that could sit on top of existing sources, centralize sports-related content, and present it in a clean, conversational, and dependable way.

Accuracy was a core requirement: the assistant had to minimize hallucinations, answer only when confidence is high, and avoid producing fabricated responses. Because the platform serves an academic audience, it also needed top-tier Croatian spelling and grammar, which was challenging given the limited availability of smaller models trained specifically for Croatian. Finally, the solution had to be fully self-hosted with complete ownership of data privacy and integrity. No user data or user prompts could ever be sent to online AI services, and every component had to operate offline without reliance on third-party systems.

Technology

Technology stack illustration

The UNIRI Sports Center AI Assistant is built on a modular, full-stack architecture. The frontend is implemented in Vue.js, with a custom design system and responsive layouts, supported by micro-interactions and smooth, mobile-first animations.

On the backend, the system combines Node.js and PHP/Laravel services with a PostgreSQL database for structured, persistent storage. AI logic is exposed through REST APIs, with dedicated modules for intent detection, routing, context handling, and response formatting. The platform is also prepared for future integrations such as wearables, sports booking APIs, and real-time messaging, making it highly extensible.

Solution

Technology stack illustration

The final solution is a smart assistant that serves as the UNIRI Sports Center's digital Information Hub. Students can instantly find facility availability, training schedules, and upcoming events, while coaches and staff access structured team overviews, session details, and clear insights into resource usage.

Behind the scenes, optimized APIs, caching, and efficient database queries ensure fast, consistent responses through a clean, animated interface designed for both mobile and desktop. The platform is built to evolve, so features like personalized training suggestions, performance metrics, or automated notifications can be added without reworking the core system. The entire solution is self-hosted on UNIRI infrastructure and aligned with strict data-privacy requirements, including the EU AI Act, and although delivered as a web application, it can be adapted into a standalone desktop app when needed.

Check out our solution live at: ai-asistent.sport.uniri.hr

Team

Technology stack illustration

Product owner

AI Developer

UI / UX Designer

Backend Developer

Frontend Developer

Let's talk about your project!