Agentic AI May End Healthcare's Payer-Provider Arms Race

Overview

The healthcare industry finds itself in a strange, counterintuitive position regarding artificial intelligence. As of 2026, a physician can deploy AI to draft a prior-authorization request, and a health plan can use separate AI tools to review it. If that request is denied, yet another system can generate an appeal.

However, the patient often remains in limbo while these automated disputes play out. We have built intelligent tools for every stage of the disagreement, but not for preventing the disagreement itself. This raises a critical question: Are we using AI to solve the wrong problem?

Providers are getting better at assembling claims and contesting denials, while payers are refining their ability to identify inconsistencies and control utilization. Yet neither side typically knows what the other requires until something goes wrong. The result is an administrative arms race where friction is automated rather than resolved.

Why This Moment Is Different

While payer-provider friction is not a new phenomenon, the conditions surrounding it have shifted dramatically. New CMS requirements are placing immediate pressure on the system, mandating shorter decision timelines for prior authorizations and requiring more specific reasoning for any denials.

By January 2027, affected payers must implement new application programming interfaces (APIs). These are intended to streamline the exchange of information among patients, providers, and health plans. Major insurers have also pledged to reduce authorization requirements, standardize electronic submissions, and move toward real-time approval decisions.

Simultaneously, AI is evolving beyond simple text generation. A new class of agentic systems can now interpret changing conditions, consult multiple data sources, and coordinate sequences of actions toward a specific goal. For the first time, regulatory pressure, interoperable data standards, and agentic AI capabilities are arriving in unison.

This convergence presents a strategic choice. Healthcare organizations can continue to use AI to help providers submit more requests while helping payers scrutinize them more efficiently. In that scenario, the administrative arms race accelerates, but the underlying adversarial relationship remains unchanged. Alternatively, stakeholders can change the target of their technological investment.

The Problem Is Not Just Paperwork

Payer-provider friction is frequently described as an administrative burden. That description is accurate, though incomplete. The deeper issue is fundamental misalignment between clinical evidence and payer requirements.

Policy criteria may only become visible to a provider after a request has already been submitted. Consequently, each organization optimizes its own part of the transaction, even when doing so creates additional work across the entire system. A denial, therefore, is not merely a failed claim. It is often a delayed signal that two organizations were operating from different information, rules, or assumptions.

Traditional Revenue Cycle Management is built to respond to the signal of friction: correct the claim, assemble the appeal, schedule the peer-to-peer review, and pursue payment. It treats friction as something to manage. A more intelligent revenue cycle would treat friction as information and move that information upstream.

Traditional revenue cycle workflows are reactive. They focus on correcting errors after they occur. Agentic AI offers the potential to shift this paradigm by treating administrative friction as actionable data, allowing organizations to address root causes before a claim is ever submitted.

Why Faster Is Not the Same as Smarter

At first glance, speed appears to be the obvious solution. If a prior authorization currently takes fourteen days, building a system that processes it in one day seems like progress. However, speed alone does not fix the underlying structural problems.

If a provider submits an incomplete request in seconds and a payer denies it in seconds, the industry has simply achieved real-time dysfunction. The transaction is faster, but the outcome remains suboptimal.

Traditional automation asks: How can I complete this task faster?

Agentic orchestration asks: Why did this task become necessary, what information is missing, and which action should happen next?

This represents a significant architectural shift. The goal is no longer to automate one isolated step, such as checking eligibility or generating an appeal. Instead, the objective is to coordinate the entire process so that the right information reaches the right system before a preventable failure occurs.

What the Architecture Actually Looks Like

The breakthrough in this space is not a swarm of chatbots sending messages to one another. It is an orchestration layer built on top of healthcare’s emerging interoperability infrastructure.

Consider a physician ordering a knee MRI in an electronic health record. The order can be represented using FHIR, a standard for exchanging healthcare data through APIs. Three implementation guides developed through the HL7 Da Vinci Project can then support different parts of the authorization process:

• Coverage Requirements Discovery (CRD): Asks the payer whether authorization is required and which coverage rules apply.
• Documentation Templates and Rules (DTR): Identifies the required evidence and helps retrieve it from the patient’s record.
• Prior Authorization Support (PAS): Packages the request, sends it to the payer, and returns the response.

These standards create the rails. Agentic AI coordinates what travels across them.

An orchestration agent could maintain the state of the request and determine which service to call next. A retrieval agent could search both structured data and unstructured clinical notes for the patient’s physical therapy history. A policy engine could compare that evidence with the payer’s current criteria. A submission agent could then assemble the required FHIR resources and send a completed authorization request.

The distinction here is crucial: the language model should help interpret messy clinical information, but it should not become the source of truth. If an agent claims that a patient completed six weeks of physical therapy, it must link that claim to the exact encounters or notes supporting it. This ensures transparency and accountability in an automated workflow.