AI in clinical practice: what recent device approvals tell us about near-term impact Artificial intelligence (AI) in healthcare is rapidly shifting from research demos to approved medical devices that assist with image analysis, triage, and workflow automation. Recent regulatory reviews and device taxonomies show image-based quantitation (radiology, dermatology lesion detection) remains a primary application, but new AI devices are also focused on automated clinical documentation, predictive analytics, and data generation for decision support. Importantly, regulatory lists and analyses highlight two themes: (1) many current approvals are narrow, specific use-cases (not general diagnostic AI), and (2) attention to generalizability and dataset bias is increasing — vendors must show performance across diverse populations. For clinics, the takeaway is pragmatic: adopt AI tools that clearly define their use, integrate safely into existing workflows, and have transparent performance metrics. Expect more ambient-capture assistants and note-generation tools to reduce clinician administrative burden in the short term.