The insurance business still has problems including slow customer service, manual document processing, and a limited application of artificial intelligence. Traditional chatbots rely on predetermined rules, lack contextual memory, and find difficulty managing lengthy documents effectively. InsuraAI combines Generative AI, Large Language Models (LLMs), and Retrieval-Augmented Generation (RAG)-based context-aware voice models to handle basic queries and deliver intelligent, personalized insurance interactions across multiple modes of interaction to address this challenge. Gathering policy data and reacting to both particular document inquiries and broad insurance questions, it makes use of Optical Character Recognition (OCR), smart query routing, and a Fine-tuned TinyLLaMA model. It enables voice communication, secure authentication using JWT or sessions, and talks with managed memory based on a LangGraph multi-agent system. Assessments highlight its precision, simplicity of usage, and ability in assisting both customers and agents. Unlike legacy RAG-based assistants, InsuraAI innovates with an agentful, multimodel design that couples document extraction, contextual retrieval and multimodal interaction to bridge a critical research gap in insurance automation.