Susil Sahu
Healthcare organizations are increasingly exploring agentic AI to improve service responsiveness, guide users, and support workflow execution. In regulated healthcare environments, however, AI agents cannot be introduced in the same way they might be used in lower-risk domains. This paper presents a governance-by-design reference architecture for agentic healthcare CRM using Salesforce Agentforce. The proposed model introduces autonomy tiers, protected health information minimization, policy-aware orchestration, human oversight boundaries, and audit-ready evidence capture as core design elements. It is intended as a practical architecture pattern for healthcare organizations that want to operationalize agentic AI in a way that is scalable, credible, and defensible.