Language-Agnostic Chatbot: A Cross-Lingual Conversational AI Framework for Multilingual Support Using Transformer-Based Architecture

Language barriers remain a significant obstacle to equitable access to information and services in an interconnected world. Traditional chatbot systems are typically designed for single-language operation, making them resource-intensive and difficult to scale.  This research presents LinguaBot, a language-agnostic chatbot framework leveraging transformer-based multilingual models such as Google Gemini API and mBERT. The system automatically detects input language, classifies intent across languages, and generates contextually appropriate responses in multiple languages. It supports over 50 languages and integrates speech recognition and text-to-speech for enhanced accessibility. Experimental evaluation shows an average intent classification accuracy of 84.3% across 15 languages and user satisfaction of 88.7%. The system demonstrates strong cross-lingual transfer capabilities while addressing challenges like low-resource languages and code-switching. LinguaBot provides a scalable foundation for inclusive conversational AI systems and future work aims to enhance cultural adaptation and expand support for underrepresented languages.