Amazon Is Giving Its Patient-Facing AI Agents More to Do - Newsweek

Direct Source Verification: This story is aggregated from Newsweek (newsweek.com). Full reporting rights and copyright belong to the primary publisher.
Amazon Web Services (AWS) is expanding its use of artificial intelligence agents in health care, introducing new capabilities designed to handle routine patient phone calls and reduce administrative workloads at clinics and health systems.

Amazon Web Services (AWS) is expanding its use of artificial intelligence agents in health care, introducing new capabilities designed to handle routine patient phone calls and reduce administrative workloads at clinics and health systems.

Naji Shafi, AWS' director of healthcare AI, broke the news today during a panel discussion at the HLTH USA conference in Las Vegas.

"Health system customers tell us up to 60 percent of [patient] calls are about routine things like booking an appointment, medication reminders or refills, getting access to the portal, hours of operation," Shafi said.
"And each of those conversations means you have to navigate a phone tree, wait on hold and do transfers to do multiple things."

"What I'm excited to announce today, right here, is we're gonna change all of that."

Amazon Connect Health can now independently manage those inbound phone calls, allowing patients to complete multiple tasks during a single conversation, even outside of traditional business hours.

The expansion comes as health systems face mounting pressure to improve patient access while managing staffing constraints. Across U.S. metros, patients wait an average of 23.5 days for a nonurgent physical with a family medicine physician, according to a 2025 AMN Healthcare survey. That's a 16 percent increase in wait times from 2009.

Dr. Will Kimbrough, chief clinical AI officer at Amazon Health Services, doesn't attribute the lag to the physician shortage alone.

"The way that primary care providers spend their time now is so slopped with extra stuff and people just doing things that may or may not be high value for them as customers or us as clinicians," Kimbrough said on Tuesday's HLTH panel, seated next to Shafi. "If we can, over the next year, build AI agents that are on the customer-facing side, that are in their pocket really helping them understand how to navigate the health care system...that's going to be worth my time and effort to do it."

The new capabilities build on Amazon Connect Health's initial launch earlier this year, which included patient verification, appointment management in preview, ambient documentation and medical coding tools.

Unlike conventional automated phone systems, the new AI agents can move between tasks, respond to interruptions and recognize when a conversation requires human assistance. For example, a patient could schedule an appointment, request a prescription refill and ask about parking without being transferred between departments.

Easier communication could improve patient satisfaction, aiding health systems in the quest to become a "provider of choice." Eighty-nine percent of patients who switch providers say ease of navigation is a reason, according to an Accenture survey cited on AWS' blog.

AWS is pointing to early results from health system partners as evidence of the technology's potential.

At UC San Diego Health, Amazon Connect Health has reduced call handling times by approximately one minute per call on a four-minute average, an AWS spokesperson told Newsweek. That translates to roughly 630 staff hours per week redirected from patient verification toward other patient needs.

The California health system has also seen a 30 percent reduction in call abandonment rates across the enterprise, with some departments reaching a 60 percent reduction, per AWS.

Temple Health, based in Philadelphia, also helped shape the expansion as an early partner and has praised the agentic call capabilities.

"Since we went live in June, Amazon Connect Health has handled patient verification on thousands of calls each week, so patients are getting help faster while our team spends less time on routine steps," said Joseph Alfonsi, AVP of patient access at Temple Health. "Building on that success, we’re excited to expand agentic patient access capabilities for prescription management, portal password resets, and information assistance."

Amazon's approach also draws on its experience operating One Medical, the primary care provider it acquired in 2023. During the HLTH discussion, Shafi and Kimbrough described how AWS engineers collaborate with One Medical's clinical teams to develop and evaluate AI tools.

"We don't bring finished, complete products to One Medical," Shafi said, explaining that AWS instead works with health care organizations as design partners to understand operational challenges. "Rather than build for health care, we build with health care."

Even so, allowing AI agents to interact directly with patients introduces questions about clinical safety, oversight and escalation.

AWS says Amazon Connect Health's agents do not provide clinical advice. Calls involving urgent concerns, patient frustration or requests for human assistance are transferred to staff.

Kimbrough described a multilayered safety approach developed for Amazon's Health AI assistant, including extensive testing, AI systems that monitor conversations for potentially unsafe responses and ongoing human clinical review; Shafi added that AWS has incorporated lessons from One Medical's development of Health AI into its own engineering approach, including deterministic rules for situations requiring human intervention.

"Trust and safety are the product," Kimbrough said.

Amazon also intends to ease the change management burden on provider organizations, Shafi told Newsweek in a November 16 interview, ahead of the official announcement. Amazon Connect Health works with existing phone systems, eliminating the need for a wholesale replacement, according to AWS. An integration with Epic allows information gathered during calls to accompany patients when they are transferred to staff, and AWS has also expanded connectivity through Redox, whose network encompasses more than 12,000 EHR and clinical data sources.

"That just reduces friction," Shafi said. "That reduces the need to learn a new thing. That reduces a cost of moving and trying a new agentic capability."

As a result, early health system adopters are recognizing value and scaling more quickly, he added.

"When we work with health systems, we really tell them start narrow and prove it out, have measurable goals, and then expand. Take Temple [Health] as an example. They started in the summer with us. Within a few weeks, they were able to scale to 1000s of calls a week," Shafi said, and now they're expanding the agents' capabilities.

"It's really transformative," he said onstage at HLTH, "and we're just getting started."

Original Source
https://www.newsweek.com/amazon-is-giving-its-patient-facing-ai-agents-more-to-do-access-health-12730024
Visit Newsweek β†—
SHARE STORY:
𝕏 f in

Related Coverage in Technology