According to a recent study by JD Power, 30 percent of commercial health plan members view their insurer as a trusted partner in their health and wellness. Health insurers are investing significantly in digital tools. A good share of those resources typically goes towards chatbots, robotic process automation and self-service portals built for easy navigation.
Where fragmented automation breaks down
Unlocking the full value of this investment has less to do with the scale of investment and more to do with the way it has been deployed. Call centers, prior authorization units, claims operations and compliance teams have automated their own functions independently, often through different vendors with no shared view of the member or case underway. Friction concentrates at the handoffs, the points where coordination matters the most.
Member services teams typically receive a steady stream of repetitive calls about claims status, eligibility and coverage; the same calls that interactive voice response systems and web portals were meant to reduce or eliminate. Prior authorization often causes friction for physicians and patients alike. A recent American Medical Association survey of 1,000 practicing physicians found that prior authorization consumes an average of 13 hours of physician and staff time each week, and 95 percent said it delays access to necessary care.
Care transitions stall too. Documentation does not always follow the patient from one system into the next. By the time compliance teams notice, they are often the ones stuck closing gaps that formed somewhere earlier in the chain.
Traditional automation was not built to do more than execute discrete tasks, and it largely performs that part reliably. It can route a call, populate a form or flag a claim for review. However, what trips it up is anything that depends on another system's data, a missing document nobody flagged yet or a judgment call about who needs to be looped in.
No single system owns the workflow, leaving members, physicians and whoever is staffing the front line to connect the pieces themselves. Insurers do not lack digital investment. They often lack coordination across investments already made.
Introducing orchestration across the workflow
Agentic AI works differently from the automation layers already in place. Instead of executing one scripted task, it can pull information from several sources, notice what is missing and chase down the follow-up needed to close the gap. However, genuine judgment calls require human oversight, with each step logged along the way. Rather than sitting within a single department, the system spans operations that otherwise wouldn’t work in tandem.
Streamlining member services and prior authorization
In member services, an AI orchestrating agent can assemble a member's eligibility status, claims history and prior interactions into a single view before a call connects. Routine cases are often resolved end-to-end. Complex ones escalate to human representatives who receive full context immediately, thereby reducing member wait times and facilitating faster processing.
Prior authorization works the same way. An AI agent checks documentation against payer criteria and initiates approved requests through connected workflows. Only ambiguous cases reach a clinician for review. This helps reduce the back-and-forth and involves less paperwork. Thus, more time is spent on the actual medical necessity decision, where a physician's judgment is critical.
Tracing claims and compliance to the same root cause
Claims denials and compliance exposure can share a common operational challenge: missing documentation paired with inconsistent handoffs between the point of care and the point of payment. As per a report by the American Hospital Association (AHA), in 2025, hospitals spent around $43 billion in total, pursuing payments from insurers for care already delivered.
On the program side, in FY 2025, the Centers for Medicare & Medicaid Services (CMS) reported an improper payment of approximately $28.83 billion for Medicare Fee-for-Service, and around $37.39 billion for Medicaid. For Medicare Advantage, most improper payments were tied to situations where supporting documentation could not substantiate the submitted information about the beneficiary’s diagnosis.
Oversight has further intensified recently: CMS initiated Risk Adjustment Data Validation (RADV) audits in 2026. This makes an audit-ready evidence trail increasingly important. An orchestration layer can capture documentation and decision evidence as work happens rather than reconstructing it after a denial or audit notice arrives.
Governing adoption and scaling deliberately
The case for agentic AI in insurance operations rests largely on how responsibly it gets deployed. Highmark Health, which operates a multistate insurance division and a 14-hospital network, is working with Abridge on a system that compares its authorization requirements with information collected during a patient visit, in real time. If some information is missing, the system prompts the physician to gather it. The collaboration is designed to accelerate approvals and support fewer denials tied to missing documentation, addressing accuracy gaps at the point of care rather than reconstructing compliance-ready records after the fact.
The same discipline needs to carry over across the industry. Escalation paths for exceptions have to be clear. Outcomes need routine monitoring for bias. Audit logs need enough detail to satisfy internal risk teams and outside examiners alike. Agentic systems can coordinate the work and gather evidence, but clinical and coverage judgment should specifically stay with humans.
The most durable path to adoption starts narrow. Insurers can deploy agentic workflows in recommendation mode on well-understood, rule-based processes, gathering documentation for prior authorization or triaging first-pass claims, where agents surface findings and humans confirm the action. As accuracy holds up against measurable outcomes, autonomy can extend into adjacent workflows within the same governance structure. This helps build trust with regulators, providers and members before agentic coordination reaches more consequential decisions.
The health insurance industry does not simply need another point solution. It needs the connective tissue that allows existing investments to collaborate. Agentic AI, deployed with disciplined governance, is positioned to serve as that key layer.
