At-Home Care Providers Face High-Stakes Learning Curves With Agentic AI 

Agentic AI is beginning to gain ground in the home-based care industry, but providers are split on how quickly to embrace it.

While some providers use agentic AI for scheduling and easing employee workloads, some experts warn that the tools can introduce immense risk to home-based care providers.

“When you go to agentic, you’re really putting your operational risk in the hands of the tool,” Greg Leighton, privacy and incident response vice chair for Polsinelli Law Firm, told Home Health Care News. “The biggest risk is that the agent will get something wrong. [For] home healthcare, that could have pretty serious consequences.”

Kansas City, Missouri-based Polsinelli Law Firm specializes in legal services relating to healthcare, including home health and hospice, as well as finance, technology and other sectors.

Agentic AI is an AI system that can make decisions, interpret goals, plan and execute actions, use tools, adapt over time and in other ways act autonomously or semi-autonomously, according to Stanford University.

Agentic AI is often considered a step above generative AI or large language models, which can generate text, images, videos and other content in response to prompts but cannot complete tasks independently.

Leighton says that there’s a larger risk with jumping straight into AI agents over generative AI or other forms of non-agentic AI.

The ramifications of an agent’s mistake depend on the task it is given. Leighton pointed to a hypothetical patient-monitoring application, where if an AI agent missed a serious medical issue or misjudged its urgency, the patient could face severe consequences or even death.

“In home health, it can get pretty high stakes pretty fast,” Leighton said.

He suggests organizations should instead evolve their existing business processes with generative AI to work through compliance issues. Implementing agentic AI workflows can come afterward.

Still, there may be a need for agentic AI in the home-based care industry. Agentic AI can help streamline call queues, Modivcare’s CEO of personal care services Marshalina Ramos told HHCN. She also noted that Modivcare is prioritizing lower hanging fruit to improve operational efficiency.

Denver-based Modivcare offers nonemergency medical transportation, personal care services, remote monitoring and integrative supportive care within 35 states.

Putting agents to work

Some providers are already exploring how agents can take on administrative work and assist employees in their daily roles.

For instance, TheKey is working toward an organizational model where employees in different roles have AI agents supporting them, Chief People Officer Emily Kraeter told HHCN.

“In our new org chart, you’ve got your leader, and then you’ve got your people, and for each of those different roles, they have you know two to three different agents that are going to be there supporting them,” Kraeter said.

Delray, Florida-based TheKey provides home care, memory care and other specialized services across 29 states and Canada, and is working to bolster its home health services.

For caregiver managers, agentic AI can augment an employee’s efficiency so that they can spend more time training caregivers, building relationships and providing support, Kraeter said. By examining employees’ workflows, the company identifies where deploying AI agents saves personnel the most time.

For instance, TheKey is rolling out Caribou’s Conductor agentic scheduling product, which the company expects to help with tasks such as finding coverage when a caregiver calls out.

What started as an isolated effort to figure out automated scheduling became the catalyst for an enterprise-wide AI strategy, Greg Meyers, chief operating officer at TheKey, told HHCN.

“Once we realized what this opened up for us, we took a step back and said every role needs this,” Meyers said. “Even my role needs some kind of AI agent that is working and counseling and looking at things across the board.”

Caution and intentionality

For providers operating in a heavily scrutinized industry, the potential benefits of agentic AI must be weighed against the risks of moving too quickly, Ramos noted.

Providers are increasingly cautious to try new things due to the amount of fraud, waste and abuse in the industry, as well as the federal government’s crackdown on illicit activity.

“We’re a highly regulated business to begin with, so nobody wants to do anything wrong or create any additional risk,” Ramos said. “People are looking for proven tech. Currently we don’t have an appetite to be the first one out to market with things.”

Providers must be deliberate in their use of agentic AI, Leighton said. Leaders should make sure they understand a tool’s business case and upsides, and how to minimize downsides.

The home-based care industry should wait for the right use cases for agentic AI, Leighton said. Some sensible use cases he sees includes streamlining operations, coordinating appointments, billing, reimbursement management and other back-office administrative tasks.

For Ramos, the central question is whether new technology can be introduced without compromising care.

“From my perspective, even as a clinician, when I think about when new pieces of technology came out, it was, ‘We’ll try it — Let’s be a little cautious though,” Ramos said. “What we do is we care for patients, and we don’t want anything to disrupt that.”