A prototype real-time Healthcare Call Agent was built using NVIDIA PersonaPlex (a research model) and LiveKit for orchestration, addressing rising call volumes, limited staff availability, and growing patient expectations for immediate support. The solution combines full-duplex speech-to-speech AI, NVIDIA GPU-powered model deployment, LiveKit’s IVR capabilities, and persona-driven prompt customization to create a virtual medical assistant available 24/7. The NVIDIA PersonaPlex real-time model integration is experimental, and its APIs, capabilities, and integration patterns may change in future releases; it is not recommended for production deployments at this time. Most traditional voice assistants rely on a cascaded pipeline of speech recognition, a language model, and text-to-speech, which can make it harder to handle interruptions, mid-sentence corrections, or overlapping speech. PersonaPlex is a full-duplex speech-to-speech research model designed for real-time conversation that can listen and speak simultaneously, enabling smoother turn-taking, faster responses, interruptions, and natural conversational backchannels. For this healthcare use case, PersonaPlex is customized with a medical assistant prompt defining the agent’s role, tone, responsibilities, and instructions for capturing patient information. PersonaPlex is built on the Moshi architecture, which supports continuous speech-to-speech conversation and uses two inputs to define conversational behavior: a voice prompt and an audio embedding.