Design and Implementation of Legal Gennie: A Retrieval- Augmented Multilingual Legal Advisory System

Legal information accessibility remains a major challenge in linguistically diverse and digitally underserved regions of India. This paper presents Legal Gennie, a retrieval-augmented, multilingual legal advisory system designed to provide simplified, context-aware legal guidance in regional Indian languages. The system integrates a Retrieval-Augmented Generation (RAG) framework with FAISS-based semantic indexing to retrieve legally grounded information from verified sources such as statutory provisions, constitutional articles, and judicial documents. Legal Gennie supports text, voice-based query input, document- aware questioning, and automated legal document generation, enabling non-expert users to interact with legal information intuitively. A multilingual processing pipeline ensures accurate bidirectional translation while preserving domain-specific legal terminology. System evaluation demonstrates high retrieval accuracy, low query latency, and strong user acceptance, particularly among rural and semi-urban participants. The results indicate that retrieval grounding significantly improves factual consistency compared to standalone generative models, while regional language support enhances legal comprehension. The proposed system contributes toward inclusive legal empowerment and demonstrates the practical applicability of RAG architectures in real-world legal advisory platforms.