Clinical Documentation & Point-of-Care AI on AWS

AWS HealthScribe sits in the documentation layer of the healthcare stack. A transcription job processes media from Amazon S3, generates a transcript file, and generates a clinical documentation file. The architecture implication is simple: HealthScribe is not the system of record for the visit, but it can become the evidence-backed note generation stage.

HealthScribe output is more than plain speech-to-text. AWS documents transcript sectioning and entities in the transcript file, and multiple template-driven note formats in the clinical documentation file. The clinical documentation guide also explains why evidence references matter: they help users validate how a summary was derived from the conversation.

Amazon Connect Health broadens the conversation from draft documentation to patient operations and point-of-care support. As of March 10, 2026, the current public materials separate feature status more precisely: patient verification and ambient documentation are generally available, while appointment management, patient insights, and medical coding remain preview capabilities.

Architecturally, Connect Health and HealthScribe can complement each other. HealthScribe is the documented API-level conversation artifact generator. Connect Health is the broader operational product surface for patient engagement, ambient workflows, and coding support integrated into care-delivery operations.

The newer Connect Health documentation also makes the control plane explicit. Teams provision a domain, create subscriptions that bind a provider or session to an agent, select templates, and decide which artifacts are vended into downstream staff workflows. That makes the service useful for point-of-care operations because outputs are not just generated once; they are routed into a governed operational workspace.