This week The Prompt Economy focuses on healthcare. It’s here that agentic AI is affecting payments, research, care protocols and the patient experience. It’s early days, but as you will see, progress is accelerating.
So is identifying the pain points in healthcare and their solutions. Health insurers have invested heavily in technology, but work still gets stuck between departments and computer systems. Infosys argues that AI agents could help close those gaps in its “Agentic AI Playbook for Healthcare Payers.” These systems can gather information, complete routine tasks and follow up when a case stalls. An agent could check coverage, collect missing documents for treatment approval or assemble the records needed to review a disputed claim. Connecting those steps could help insurers resolve requests faster and reduce repeated calls from members. Agents could also track follow-up care after a hospital discharge and flag missed appointments.
Infosys says those gains will depend on how insurers organize the work and supervise the technology. Insurers should start with a focused task, connect the relevant data and keep people responsible for decisions that require clinical judgment. Agents can initially recommend actions while human reviewers make the final call. Insurers shouldn’t expand their authority until accuracy, fairness and security are proven. The systems should also record their actions and the reasons behind them, helping compliance teams spot problems earlier.
Infosys notes that most member-service deployments remain in pilot or early implementation stages, making measured expansion central to its approach.
Reducing Staff Workload With In-House AI
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
AI agents are helping New York’s Hospital for Special Surgery reduce the time staff spend on insurance claims, according to an article from MIT Technology Review Insights, produced in partnership with AI developer Ema. The hospital’s chief digital and technology officer, Dr. Ashis Barad, says agents now complete 1,100 claims a month. They’ve cut the appeals stage from 45 minutes to five and raised the appeal success rate from 65% to 100% over nine months. The hospital now handles all claims in-house. It’s also deploying an AI service with Ema that helps patients find and book the right clinician. The service considers symptoms, location, insurance coverage and appointment availability.
The article argues that these systems could give healthcare workers more time for patients by taking on administrative work. But providers need connected records, consistent information and clear rules for human oversight. At Hospital for Special Surgery, the scheduling and triage service sends sensitive or uncertain cases to human specialists. Staff can review its decisions and intervene at any point. An internal committee also reviews AI projects, with closer scrutiny for systems that could affect patient care.
Barad’s broader goal is to make the technology available across the hospital, supported by staff training. The potential benefit is more time for the complex cases and personal attention that patients need.
Agentic AI Makes Inroads for Elder Care
One of the more interesting use cases is from Australia. Wesley Mission has deployed three AI agents to reduce administrative work in aged care, according to Health Services Daily. One helps with care plan assessments, saving staff between 30 minutes and an hour per shift. Another has cut the daily task of verifying visits from six hours to about 15 minutes. A third handles calls about sick leave and canceled shifts, finds replacement workers and recommends a staffing change to a supervisor.
Wesley Mission executive Scott Shipman says the aim is to give staff more time with clients. All three agents require human approval before taking final action.
The experience also offers a practical lesson for healthcare leaders: fix the underlying process before adding AI. Shipman described one agent that appeared to be malfunctioning. The problem turned out to be a staff member who wasn’t following the process, leaving the agent stuck repeating steps. Speakers at the Health Services Daily webinar advised hospitals to review the AI tools employees already use and establish clear responsibility for oversight. They also recommended starting with a limited task that can show measurable improvements within weeks. Staff training, ongoing monitoring and human approval can help organizations build confidence before giving agents broader responsibilities.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
Add as Preferred Source
