A lack of formal clinical training often limits the effectiveness of community health workers (CHWs) and, thus, their ability to generate timely and accurate insights based on the information shared with them. Integrating large language models (LLMs) into the CHWs’ workflow presents a novel opportunity to enhance their decision-making capabilities by providing real-time, context-specific, and evidence-based guidance. This non-interventional study aims to explore whether LLMs can support more effective decision-making. After an interaction between a CHW and a patient, the LLM will analyze the recorded conversation to determine the appropriateness of referral decisions made by CHWs compared to the LLM against the clinical expert panel consensus. The study also assesses the feasibility and appropriateness of an LLM-based clinical decision support system for CHWs. The results will inform future research, including the need for an interventional trial of such a product.