Announcing agentic AI for healthcare patient engagement in Amazon Connect (Preview)

We’ve all been there: waiting on hold for what feels like an eternity just to schedule a routine doctor’s visit or repeating the same information to multiple people before finally reaching someone who can help. These frustrating experiences aren’t just inconvenient for patients—they’re also an increasingly large burden on health systems and their employees.

Healthcare organizations are drowning under the weight of patient communication demands. Large health systems now field millions of patient calls annually while managing an ever-increasing volume of messages and emails. When routine interactions (such as booking a doctor’s appointment) become obstacles, patient loyalty erodes quickly. A staggering 89% of patients cite care navigation challenges as their primary reason for switching healthcare providers. This creates additional pressure on healthcare providers, who are already stretched thin.

The scale and cost of patient engagement have become unsustainable, stemming from a combination of fragmented systems and tools, and outdated processes for routine tasks like scheduling, refills, referrals, and prior authorizations. These inefficiencies are burdening healthcare staff and forcing patients to navigate complex administrative networks, while managing their health conditions.

Amazon Web Services (AWS) is excited to announce the preview of new purpose-built agentic AI capabilities in Amazon Connect to help healthcare providers scale patient verification and appointment management without requiring complex customization from overworked IT teams. Unlike traditional deterministic workflows, or chatbots that follow rigid scripts, our agentic AI capabilities can reason, plan, and take autonomous actions on behalf of patients. They can reason through complex, multi-step workflows: from provider discovery through appointment confirmation, while adapting to unique situations and making intelligent decisions about when to escalate to human staff.

These capabilities feature secure, real-time integration with electronic health records (EHRs), enabling self-service verification for patients and caregivers, and verifying appointments are scheduled with up-to-date, accurate information. The appointment management AI agent handles appointment-related calls, which represent on average 50% of total patient call volume, eliminating manual scheduling tasks and freeing staff to take on high-value, specialized patient interactions.

Enterprise-grade AI built for healthcare

Building reliable AI for healthcare means performing predictably across millions of patient interactions, while navigating unique requirements that don’t exist in other industries. It needs to consider patient safety protocols requiring immediate escalation, regulatory compliance like HIPAA with comprehensive audit trails, and real-time EHR integration. It should also handle multi-factor identity verification, insurance validation workflows, and have a contextual understanding of medical terminology. This is all while preserving patient trust during vulnerable moments.

EHRs serve as the central source of truth for all clinical records and patient workflows in modern healthcare systems. The healthcare AI capabilities in Amazon Connect are designed with this understanding, integrating directly with the EHR system to verify that all patient interactions are based on the most current and accurate information. Rather than creating a separate system that requires synchronization, our solution works as an extension of the EHR, maintaining data integrity while enhancing accessibility through conversational AI.

Amazon Connect is trusted by leading healthcare organizations like NHS Midlands and Lancashire (NHS ML) and Tufts Medicine for patient engagement. Feedback from these real-world implementations has informed our new HIPAA-eligible healthcare AI capabilities, which are purpose-built to address critical patient communication challenges through specialized architecture. The appointment management agent integrates directly with the EHR, delivering healthcare-specific guardrails, contextual escalation protocols, and complete management of the appointment scheduling task—from insurance verification to provider matching based on patient history (Figure 1).

Figure 1: Appointment management AI agent workflow

Safety-first design with comprehensive guardrails

The AI safety architecture extends far beyond standard chatbot protections, incorporating multiple layers of content filtering and behavioral safeguards designed specifically for healthcare. We have developed and rigorously tested comprehensive escalation protocols spanning patient safety, clinical needs, and patient experience. These protocols are continuously refined through real-world testing and feedback to confirm the AI agent responds with empathy and provides appropriate handoffs to human staff when specialized care is needed.

Behavioral safeguards:

• Medical concern detection with immediate escalation to clinical staff
• Patient frustration monitoring with compassionate handoff protocols
• Communication barrier assistance for language or accessibility needs
• Specialized care needs recognition for scenarios requiring human expertise

Content protection:

• Personally identifiable information (PII) detection and blocking to prevent unauthorized data exposure
• Prompt injection prevention to maintain conversation integrity
• Medical advice restrictions with immediate clinical escalation protocols
• System architecture protection preventing unauthorized access attempts

For example, if after booking an appointment, a patient requests wheelchair accessibility for the exam room, the appointment management agent recognizes this specialized need and immediately escalates to human staff (Figure 2). The AI agent provides an intelligent, contextual summary that highlights the important points the human staff needs for seamless handoffs, including patient intent, preferences, verification status, and conversation context. This eliminates the need for patients to repeat information.

Figure 2: Example of intelligent call handoff in Amazon Connect

Enterprise security and compliance

The healthcare capabilities in Amazon Connect are engineered to handle millions of patient interactions while maintaining the highest standards of healthcare data protection, privacy, and regulatory compliance. The security architecture is built specifically for healthcare’s unique requirements and scales seamlessly across the organization.

Requirements include:

• HIPAA-eligible infrastructure with comprehensive audit trails
• Zero-persistence architecture that retrieves EHR data in real-time without storing patient information, ensuring data remains only in the EHR system
• Secure EHR APIs and end-to-end encrypted connections for real-time EHR data access
• Role-based access controls and identity verification for all system interactions
• Continuous monitoring and performance measurement to facilitate consistent service quality

As part of Amazon Connect, a solution built to power the customer service operations of Amazon along with tens of thousands of customers worldwide serving millions of customers daily, these healthcare capabilities can handle the most demanding patient engagement workloads with enterprise reliability and security. We continuously update our security architecture based on best practices, verifying healthcare organizations can trust that their patient engagement is built on the latest security standards. As healthcare regulations evolve and technology advances, Amazon Connect adapts to meet new compliance requirements without compromising performance or patient experience.

Transforming patient engagement at scale

With Amazon Connect, AI works together with human staff to facilitate a personalized patient experience. The AI agent can deliver fast resolution for routine tasks (like appointment scheduling), with contextual escalation capabilities and real-time patient summaries for moments requiring human support. This seamless collaboration maintains strict governance and compliance, so healthcare teams can focus on specialized patient interactions.

Customers like UC San Diego Health are using Amazon Connect to improve patient engagement and empower their staff to focus on the moments that matter the most. On December 4, 2025, at AWS re:Invent, UC San Diego Health will share how they have implemented healthcare AI capabilities in Amazon Connect. Join their session in-person or on demand after the event.

Getting started

The preview of healthcare agentic AI capabilities in Amazon Connect is currently available for US customers. To begin using these healthcare AI capabilities in Amazon Connect, request access through your AWS account team or contact an AWS Representative.