Neurological conditions such as ALS create complex communication challenges that evolve as patient abilities decline. This session breaks down how to architect an adaptive GenAI communication platform that remains reliable, safe, and personalized over time. We’ll cover a modular system design combining multimodal inputs (eye-tracking, head pose, EMG, BCI) with a vectorized personal knowledge base to generate context-aware communication and intent. You’ll learn how supervised agentic workflows, retrieval-augmented generation, and dynamic capability adaptation enable fast, relevant responses while preserving a user’s tone and identity. We’ll share key engineering insights from clinical deployments, including performance metrics such as information transfer rate (ITR), communication efficiency, and usability (SUS).