Effective interpretation of patient-provider communication across secure messaging, written documentation, phone calls, and video interviews is essential for informed clinical decision-making, including critical processes such as trial eligibility assessment. However, analyzing this unstructured, multimodal data at scale is time-intensive and demands expert clinical insight, often overwhelming healthcare professionals. To address these challenges, we present a novel Agentic AI framework: a modular, multi-agent system designed to augment and streamline qualitative analysis in healthcare settings. Our framework integrates reasoning-capable large language models (LLMs) with structured medical knowledge drawn from domain-specific knowledge graphs, enabling specialized agents to perform tasks such as clinical coding, thematic analysis, bias detection, and automated report generation. Coordinated through a LangGraph-style orchestration layer with shared memory, these agents deliver context-aware, adaptive, and error-resilient workflows tailored to clinical needs. By automating labor-intensive qualitative tasks and embedding transparency and oversight mechanisms, this system may significantly reduces clinician workload and accelerates the interpretation of complex patient narratives. Evaluations on three real-world, open-source healthcare datasets demonstrate strong alignment with expert interpretation and reveal nuanced insights often overlooked in manual review. Ultimately, our approach empowers Subject Matter Experts (SMEs) to focus on higher-level critical thinking and decision-making, potentially improving patient outcomes and clinical trial efficiency.