Curtis is a contract manager, four years in, and he is good at this. He reads the denial letter twice before he says anything, which is more than most people do.

The medical necessity determination cites a criterion. The criterion is real, it appears in the plan's published policy, and it does not fit the case. Not marginally. The patient's documented presentation does not resemble the situation the criterion describes.

His first instinct is the correct one for the world he trained in. Somebody misread the chart. Escalate, request a peer to peer, get a physician on the phone, and the error corrects.

Then he notices the timestamp. The determination came back in eleven minutes.

No one misread the chart, because nobody read the chart.

Every agreement in your portfolio assumes a person

Go pull your largest commercial contract and read the utilization management article. Then read the appeals language. Then read the definitions.

Somewhere in there you will find the assumption. Medical necessity determinations are described in terms that presume a human clinician applying judgment to a record. The appeal structure is built around persuading that clinician. The peer to peer provision exists because the drafters believed there was a peer.

That assumption was accurate when the language was written. It is not reliably accurate now, and almost no contract in circulation has caught up.

This is not a claim that automation is illegitimate. Plenty of automated adjudication is appropriate, faster than a human, and better for everybody including the patient. The problem is narrower and more specific. Your contractual rights were built around a process that has changed, and nothing in the agreement gives you standing to ask what the new process is.

Regulation is arriving, and it is not going to solve this for you

States moved substantially on this in 2026. Legislatures have been enacting laws governing insurer use of artificial intelligence in prior authorization and claims decisions, adding transparency requirements and human oversight mandates for adverse determinations. It is genuine movement and it matters.

It is also uneven, it is state by state, and it does not run in one direction. Georgia enacted a law, effective January 2027, that expressly permits health insurers to use AI in the prior authorization process, including participation in decision-making. Read across the full set and what you have is a patchwork where some states constrain the practice, some states authorize it, and most of your book of business is governed by whichever set of rules happens to apply to that product in that state.

Here is what none of it does. It does not give you a contractual right. It gives you a regulatory complaint, filed after the fact, through an agency, on a timeline that has nothing to do with your revenue cycle.

There is a difference between a rule a plan must follow and a term a plan owes you. The first is enforced by somebody else, eventually, at their discretion. The second is enforced by you, at renewal, with a remedy you negotiated.

Waiting for regulation to produce the second one is a strategy that has never worked in managed care and is not going to start now.

Six clauses your contract does not have

This is the priority list I would put in front of your general counsel before your next commercial or Medicare Advantage renewal. None of these ask a plan to stop using automation. Every one of them assumes it will.

(1) Disclosure. The plan discloses on the determination letter when an automated or algorithmic process contributed to an adverse determination. Not the vendor, not the architecture, not anything proprietary. The fact of it. This is the foundational clause because every other one is unenforceable without it.

(2) Human review attestation. No adverse medical necessity determination issues without documented review by a licensed clinician, with the reviewer's credentials and specialty stated on the letter. This is the clause with the most regulatory tailwind behind it and it is the easiest to argue for, because plans generally assert they already do it.

(3) Change notice. Material changes to automated medical policy engines and claim edit logic receive the same advance written notice the contract already requires for fee schedule and medical policy changes. This is a symmetry argument. You already have the provision. You are asking that it cover the mechanism that now does the work the provision was written about.

(4) Audit right with a defined sample. You may request, on a stated number of determinations per quarter, the specific criteria applied and whether an automated process contributed. Bounded, not open-ended. Bounded asks get agreed to.

(5) Timeline parity. Automation does not become a one-way speed advantage. If the plan can issue an adverse determination in eleven minutes, your response window does not shrink, and an automated denial triggers expedited appeal handling. The principle is simple: speed on their side does not create burden on yours.

(6) A remedy. The clause that makes the other five real. A defined consequence when one through five is breached, whether that is automatic overturn, interest, or a reopener trigger. A right without a remedy is a preference.

If this is landing for you, subscribe to Mastering Managed Care and follow me on LinkedIn: Kevin W. Barron, FHFMA, FACHE. This is where I post what we're all working through in real life.

Why this one is different from every other structural ask

Most structural terms are a fight over money wearing a procedural costume, and both sides know it. This set is not, and that changes the conversation.

You are not asking the plan to pay more. You are not asking it to approve more. You are asking it to tell you how a decision was made and to stand behind the decision the same way it would have five years ago. A plan that is doing this responsibly loses nothing by agreeing. That is an unusual position to negotiate from and it is worth using while it lasts.

It will not last. Once these clauses become standard, they will be priced like everything else that became standard. Right now they are novel enough that most network executives do not have a prepared position on them, which is the single best moment to introduce a term.

What happens if you skip it

The volume of automated adverse determinations is going to keep rising, because the economics are overwhelming and the technology works. Your appeal function scales linearly, staffed by humans, against a determination process that does not scale linearly at all.

That is the asymmetry. It is not primarily about whether any individual denial was right. It is that you will be answering an accelerating process with a fixed-capacity one, and the gap between them is the amount of legitimate revenue you write off without ever contesting.

Meanwhile your contract will still describe a peer-to-peer conversation with a physician who considered the record.

What the other version looks like

Curtis gets that eleven minute determination and the disclosure clause means the letter already tells him an automated process contributed. He does not spend three days trying to reach a reviewer who was never there. He routes it directly into the expedited path the contract created for exactly this, cites the human review attestation clause, and gets an actual clinician on the record.

The appeal takes hours instead of weeks. More importantly, the pattern becomes visible. Ninety days of disclosed automated determinations, sorted by service line, is a document. And a document is the thing you take to the joint operating committee.

None of that requires the plan to slow down. It requires the contract to acknowledge what is actually happening.

Do this

Pull the utilization management and appeals articles from your three largest agreements this week and find every place the language assumes a human being made the decision. That list is your redline for the next renewal.

The party across the table is still a company. The party on your denial letter increasingly is not. Your contract should be written to both.

About the Author

Kevin W. Barron, MBA, FHFMA, FACHE is a nationally recognized healthcare contracting and payer relations executive with over 30 years of experience in healthcare finance. He is an author, speaker, and mentor to emerging leaders in healthcare finance. Follow Kevin on LinkedIn or subscribe to his weekly newsletter at https://www.KevinWatsonBarron.com

Disclaimer: This article is provided for general informational and educational purposes only. It reflects my personal views and observations based on professional experience in healthcare finance, managed care, payer relations, and revenue cycle operations. It should not be interpreted as legal, financial, regulatory, actuarial, or reimbursement advice. The content is not intended to represent the official position of my employer, any payer, provider, professional association, or other organization with which I may be affiliated. Readers should consult their own legal, financial, compliance, actuarial, or operational advisors before making decisions based on the issues discussed. Any references to payers, providers, regulations, market trends, or reimbursement practices are intended for discussion and education only and should not be construed as a statement about any specific contract, negotiation, patient matter, or confidential business arrangement.