Amazon Web Services is pushing deeper into agentic artificial intelligence for healthcare, taking aim at common administrative tasks such as scheduling, medical histories, clinical documentation and coding.
The agentic AI solution, Amazon Connect Health, is the first purpose-built AWS solution for healthcare providers, the company announced Thursday. Developed for healthcare providers and health technology companies, Amazon Connect Health offers five agentic AI capabilities that use natural language voice tech to support administrative work like patient verification and appointment scheduling for call center staff while also cutting down on manual work for clinicians by synthesizing patient medical histories and leveraging ambient tech to transcribe medical visits. The solution also streamlines medical coding after the patient-provider visit.
"In conversations with large health systems, AWS has found that staff spend up to 80% of call time on manual data compilation across fragmented tools. Verifying patient identities, scheduling appointments, manually stitching together patients' medical histories scattered across multiple systems, meeting complex documentation requirements: these tasks pull clinicians and their teams away from patient care. 89% of patients said care navigation challenges—difficulty scheduling, long wait times, and access barriers—were their reason for switching providers," Colleen Aubrey, AWS' senior vice president of applied AI solutions, wrote in a blog post.
The AWS solution be deployed in days instead of months with native electronic health record and Amazon Connect contact center integrations, prebuilt connectivity to more than 100 EHR via data integrator partners and a fully managed unified SDK, according to the company.
If a patient calls a provider and says, "I want to see my doctor after work next week." Amazon Connect Health understands the reason for the call, the patient's context and their preferences, according to Aubrey. "It confirms who the patient is, checks their insurance, reviews patient and provider availability—and books the appointment while the patient is still on the line. Amazon Connect Health combines the power of Connect, AWS's AI-powered customer experience solution, with real-time connection to the EHR," Aubrey wrote.
When a situation requires a human touch, such as a medical concern or a complex request, the solution escalates seamlessly to staff. Health systems can customize exactly when and how that handoff happens, according to AWS.
Some health systems have already deployed capabilities from Amazon Connect Health. UC San Diego Health, which handles 3.2 million patient interactions annually, reports it is saving one minute per call, diverting 630 hours weekly from patient verification to direct patient assistance and reducing call abandonment rates by 30%—as high as 60% in some departments, according to the blog post.
Beyond appointment scheduling, AWS built the solution to review a patient's medical history before a doctor visit to provide the clinician a concise summary and insights, including active conditions, recent events, trends over time and chronic conditions that may be relevant to both care gap closure and accurate billing.
AWS expanded its ambient scribe capabilities to transcribe the doctor-patient conversation and draft clinical notes for provider review in real time. The solution also generates patient-friendly after-visit summaries for provider review and the medical codes clinicians need for billing, with each code linked to source evidence for auditing.
Amazon One Medical clinics have been testing out the solution, and the ambient documentation now spans more than a million visits, with strong clinician adoption and regular weekly usage, Aubrey wrote. Amazon One Medical plans to expand to intelligent medical coding this year.
Netsmart, an EHR company that serves community-based providers with more than 1,300 client organizations, also has integrated Amazon Connect Health and reports ambient documentation adoption increased by 275%.
AWS also makes the case for its stringent security, patient privacy and data integrity standards. Aubrey notes in the blog post that AWS has 130 HIPAA-eligible services and certifications for global IT and compliance standards.
The company uses what it calls evidence mapping—a feature that links every piece of AI-generated output back to its exact source, "whether that's an ambient conversation transcript, a patient's medical records or billing guidelines," Aubrey wrote.
"If an AI-generated summary says 'patient reports poor diet,' the doctor can click to hear the exact moment in the conversation where that was discussed. This transparency supports faster, safer review—so clinicians can audit, refine and finalize with confidence," Aubrey wrote.
Amazon Connect Health leverages specialized supervised fine-tuning and reinforcement learning techniques on healthcare specific datasets and guidelines. The AI capabilities undergo multistep evaluation of model performance for safety and accuracy. This includes LLM-as-judge based evaluations, where one AI system checks another's work, as well as clinician-in-the-loop driven checks, Aubrey wrote.