Designing culturally aware health chatbots for underserved communities is essential to advancing global health equity. A major challenge, however, is the conversion of contextual, often qualitative user requirements into technical specifications. In this paper, we present a natural language processing (NLP) pipeline for the co-design of culturally aware health chatbots. The pipeline systematically converts user stories into system specifications. By employing Large Language Models (LLMs), the pipeline automates the extraction of cultural and contextual requirements from user stories, which inherently counteracts biases from general AI models. By using the requirements engineering phase to direct attention to domainspecific and culturally relevant data, this approach ensures that the resulting specifications for chatbots are grounded in the realities of the communities they aim to serve. This approach does not only improve the efficiency of the design process, it also proactively embeds cultural sensitivity and inclusivity at the heart of AI-driven health solutions