AI agents in healthcare: Improving patient support while ensuring privacy

The integration of AI agents into healthcare systems can transform patient engagement, streamline operations, and reduce administrative burdens. This paper explores the application of Large Language Model (LLM)-powered AI agents in patient-facing use cases such as appointment scheduling, benefits navigation, and claims assistance. We introduce a framework that balances conversational efficiency with rigorous adherence to patient privacy, leveraging retrieval-augmented generation (RAG) grounded on HIPAA-compliant knowledge sources. We evaluate privacy-preserving strategies, including selective data masking, tenant isolation, and zero-retention prompt flows. We demonstrate measurable improvements in support resolution time and patient satisfaction through enterprise case studies while maintaining strong data governance. This work contributes to the growing need for ethically responsible AI in healthcare and sets a precedent for deploying trustworthy agents in regulated environments.