A regional healthcare payer was looking for a creative solution to solving one of the most common friction points for their members: the process of finding an in-network provider. Traditional provider search tools are often rigid, keyword-dependent, and difficult to navigate, leading to frustration, abandoned sessions, and an over-reliance on live agents for assistance.

To demonstrate the potential of modern large language models (LLMs), we rapidly developed a proof-of-concept

mobile application using React Native. The application leveraged GPT-4o, enriched with contextual knowledge about the user including their provider search history, home and work locations, and persona

lized preferences to deliver intelligent, conversational guidance.

We also implemented a model context protocol server that dynamically interfaced with provider search APIs, tailoring responses in real time based on user interactions. This allowed members to find relevant providers with minimal input and receive actionable next steps such as initiating a call to the provider’s office or booking an appointment directly through platforms like ZocDoc.

OpenAI API, React Native, Typescript, Model Context Protocol (MCP)