Access to accurate health information in India is often limited by language barriers, digital divides, and a shortage of qualifiedhealthcare professionals at the Primary Health Centers (PHCs) and community awareness hubs. This paper proposes thedeployment of localized Large Language Models (LLMs) to enable multilingual, conversational health communication forpatients and caregivers. The solution combines interactive, low-cost kiosks installed at PHCs and public health awareness centers with a web-based application for at-home access, allowing individuals to inquire about common diseases, symptoms, preventivecare, and treatment guidance in their native languages. The system is designed to operate offline using locally hosted LLMs inareas with poor internet connectivity, while seamlessly switching to cloud APIs when connectivity is available. By bridginglanguage and access gaps, this approach aims to enhance early disease awareness, patient engagement, and healthcare equity,supporting the government’s vision for digitally inclusive primary healthcare in India. The paper discusses the architecture,hardware specifications for kiosks, language model selection and fine-tuning strategies, as well as privacy and data securityconsiderations necessary for safe and responsible AI deployment in public health environments