AWS launches the first purpose-built AI agent suite for the medical frontline - ‘Amazon Connect Health’

Last week, Amazon Web Services (AWS) officially launched Amazon Connect Health in a perfectly timed move ahead of the upcoming HIMSS healthcare conference. This is the company's first industry-specific "agentic" platform, designed to act as a digital teammate that can autonomously execute complex clinical and administrative tasks that typically "drown" healthcare workers in paperwork.

The suite is already being deployed by major providers including UC San Diego Health, One Medical, and Netsmart, with initial data showing it can save staff hundreds of hours per week by automating the "three-call marathon" of patient navigation.

From ‘chat’ to ‘action’: The five core agents

While previous AI tools focused on summarizing text, Amazon Connect Health features agents capable of "planning and reasoning" to complete full workflows.

• Patient verification (Available Now): Uses natural language to verify identities and insurance in real-time, integrating directly with Electronic Health Records (EHRs) like Epic.

• Ambient documentation (Available Now): An evolution of AWS HealthScribe, this agent listens to doctor-patient visits, transcribes them, and automatically populates clinical notes and after-visit summaries.

• Appointment management (Preview): A 24/7 natural language voice agent that can check a doctor's actual availability, verify prior authorizations, and book the slot without putting the patient on hold.

• Clinical summaries (Preview): Before a doctor enters the room, this agent scans fragmented medical histories across different systems to provide a "pre-visit brief" of chronic conditions and recent labs.

• Medical coding (Coming Soon): Automatically suggests the correct billing and diagnosis codes based on the documented visit, aiming to accelerate the revenue cycle for hospitals.

The ‘sovereign data’ foundation

To address the intense privacy requirements of the medical field, AWS is tying these agents to a newly upgraded data layer.

The agents run on top of AWS HealthLake, which now includes a "Data Transformation Agent." This tool can take legacy, messy medical files (CCDA) and convert them into modern, queryable formats (FHIR) in days rather than months.

To build trust with skeptical clinicians, every AI-generated summary or code includes "evidence mapping". That’s clickable links that show exactly which part of the transcript or medical record the AI used to reach its conclusion.

Additionally, AWS is using a secondary, highly-calibrated AI model to "critique" the primary agents, ensuring outputs remain accurate and clinically safe.

The $99 ‘teammate’ business model

AWS is moving away from complex "per-token" cloud pricing toward a more predictable software-as-a-service (SaaS) model for this platform.

The service is priced at $99 per user per month, covering up to 600 patient encounters.

With 89% of patients reporting "care navigation" problems and clinicians spending up to 80% of their time on manual data entry, AWS is positioning this as a "retention tool" for overstretched nursing and administrative staff.

"Our healthcare workers are drowning in administrative complexity," said Naji Shafi, General Manager of Health AI at AWS. "We’re moving from AI that helps you do the work to AI that actually does the work, acting as a teammate instead of just a tool."