AI agents in clinical practice: an evidence map

Abstract The rapid emergence of autonomous AI systems (agentic AI) marks a pivotal inflection point in digital medicine. This perspective synthesizes evidence across diagnostics, management, and care-adjacent operations to identify opportunities for adoption. Early deployments are concentrated in administrative workflows, though agentic applications are rapidly expanding across the clinical patient journey. Proactive governance, auditability, and clinician oversight are essential for responsible agent translation into clinical practice.