CMS WISeR Model Is Live: What Medical Coders Must Know in 2026

CMS WISeR Model Is Live: What Medical Coders Must Know in 2026

On January 1, 2026, CMS quietly flipped the switch on one of the most significant prior authorization experiments in Medicare history. The Wasteful and Inappropriate Service Reduction (WISeR) Model -- run by the CMS Innovation Center (CMMI) -- handed prior authorization decisions for select Medicare procedures to six private AI vendors operating across six states. If your facility submits claims in Texas, New Jersey, Oklahoma, Ohio, Washington, or Arizona, your documentation is now being reviewed by an algorithm before a human ever sees it.

For medical coders, WISeR is not an abstract policy debate. It is a live workflow change that directly affects whether certain procedure requests are approved, delayed, or denied -- and coding specificity is the deciding variable.

What Procedures WISeR Covers

WISeR targets a defined set of services CMS has identified as high-cost and potentially overutilized in traditional Medicare. The covered service categories include skin and tissue substitute procedures, spinal cord stimulators and other neurostimulator implants, and knee arthroscopy. These are not obscure edge cases. Spinal neurostimulator procedures, for example, carry CPT codes in the 63650--63688 range and are among the more expensive outpatient procedures Medicare processes each year.

The model runs for six performance years through December 31, 2031, in its assigned states. Vendors are paid based on how much they save Medicare, adjusted for quality metrics including provider experience scores -- meaning the financial incentive is to approve quickly when criteria are met and deny when they are not, with minimal middle ground.

The Six AI Vendors Handling Reviews

CMS selected six vendors in November 2025, each assigned to one state. Understanding what their platforms actually do tells coders a great deal about what documentation will and will not satisfy review.

Cohere Health (Texas) operates the Cohere Unify platform, which handles over 12 million prior authorization requests annually for 660,000-plus providers. Its AI delivers real-time nudges to providers flagging missing information before submission -- meaning incomplete or unspecified codes that trigger a hold will surface before the claim is even filed. Innovaccer (Ohio) launched Flow Auth in August 2025, an end-to-end agentic AI system that automatically detects prior auth requirements, builds payer-ready clinical packets, submits requests, and -- critically -- drafts automated appeals. That last feature is a signal: the system is built around the assumption that some requests will be denied on first submission.

Genzeon (New Jersey) is the most transparent about its layered approach: robotic process automation handles intake and system updates, while agentic AI takes over for decision support -- summarizing clinical records, comparing documentation against clinical guidelines, and triaging by complexity. Humata Health (Oklahoma), founded by former Mayo Clinic radiologist Jeremy Friese, MD, positions its platform as "built for yes," targeting 90% touchless approvals while guaranteeing the system never auto-denies -- complex cases always get human review. Virtix Health (Washington) focuses on clinical data connectivity and risk adjustment coding validation through its LINX Platform. Zyter|TruCare (Arizona) integrated agentic AI into prior authorization intake in mid-2025, combining optical character recognition and automated routing to minimize manual steps.

The common thread: every platform ingests ICD-10-CM and CPT codes from the clinical record and cross-references them against coverage criteria in real time. Vague or unspecified codes do not generate the clinical signals these systems need to auto-approve.

What Coders Must Do Differently

ICD-10-CM Specificity Is the Gating Factor

WISeR procedures are high-specificity by nature. A spinal cord stimulator implant for chronic pain requires the coder to capture the specific spinal region (cervical, thoracic, lumbar), laterality where applicable, and the underlying diagnosis at its most granular ICD-10-CM level. Codes like M54.5 (low back pain, unspecified) that once passed through payer systems without friction will now fail to satisfy AI-driven criteria matching. The WISeR-assigned AI platforms are comparing diagnoses against coverage guidelines that require documented clinical indicators: failed conservative treatment, specific duration of symptoms, and functional impairment scores.

For knee arthroscopy, the relevant diagnoses must specify whether the condition is medial or lateral, traumatic or degenerative, and whether imaging confirmation exists in the record. AI platforms that ingest clinical documentation will look for these specifics whether or not the coder has captured them -- and if the code is unspecified, the system flags the case for human review or denial rather than auto-approval.

CPT Code Selection Determines Coverage Mapping

For skin and tissue substitutes, the specific HCPCS product code matters as much as the CPT application code. Different graft products carry different coverage criteria. Coders selecting a non-specific application code without the correct HCPCS product code will generate a documentation mismatch that AI review systems will not resolve in the provider's favor. Facilities whose coding workflow does not link the product code to the procedure code will see disproportionate prior auth delays starting in 2026.

Five Things Coders Should Do Now

• Map your high-volume WISeR procedures. Identify all claims in the covered service categories submitted to Medicare in the six WISeR states and flag any that regularly use unspecified or less-specific diagnosis codes.

• Update code pairing edits. Build or audit your chargemaster and encoder pairing rules for spinal neurostimulator, tissue substitute, and knee arthroscopy procedures to ensure CPT and HCPCS codes are linked correctly.

• Coordinate with clinical staff on documentation specificity. The documentation gap that creates coding gaps starts in the clinical note. CDI outreach to surgeons and proceduralists ordering WISeR-covered services is a practical first step.

• Monitor denials by vendor geography. Because each vendor operates in one state, denial pattern analysis by state will quickly reveal whether one vendor's criteria interpretation is tighter than others -- intelligence that can inform appeals strategies.

• Prepare denial appeal workflows early. Innovaccer's Flow Auth auto-generates appeals. Your facility's appeal process should be at least as fast. Have a pre-built appeal template with ICD-10 and CPT documentation requirements for each WISeR service category ready before the first denial arrives.

The Controversy Coders Should Understand

WISeR did not arrive without opposition. Before the model launched, a bipartisan group of lawmakers introduced legislation to cancel it, arguing that AI-driven Medicare prior authorization would limit patient access to care. CMS proceeded anyway, and the model is now in its first performance year. Humata Health's explicit pledge -- that its system never automatically denies requests, only auto-approves those that clearly meet criteria -- is a meaningful distinction. But it also means that any request that does not clearly meet criteria goes to human review, which introduces delay. For coders, the practical implication is the same either way: documentation precision reduces the probability of landing in the human-review queue.

The controversy also underscores something important for coders to communicate upward: WISeR is not a permanent policy shift on its own, but it is a six-year pilot with financial incentives structured to reward savings. If it demonstrates savings without measurable harm to access, it is likely to expand. The coding implications described here are likely to get more consequential, not less.

The Bottom Line

The CMS WISeR Model puts AI vendors in the prior authorization seat for Medicare procedures that many facilities bill frequently. The AI platforms being used -- from Cohere Health's real-time nudge system in Texas to Innovaccer's multi-agent Flow Auth pipeline in Ohio -- are designed to make rapid decisions based on the clinical evidence coded into the claim. Specificity in ICD-10-CM diagnosis codes and accuracy in CPT and HCPCS procedure code pairing are no longer just accuracy metrics. In WISeR-covered states, they are approval criteria.

Facilities that close the documentation-to-coding gap on spinal stimulators, tissue substitutes, and knee arthroscopy before the first batch of denials arrive will recover revenue faster than those that treat WISeR as a future concern. The model is already live. Coding is the lever providers still control.

For practices looking to get ahead of AI-driven payer scrutiny across all procedure types, Medikode's automated medical coding platform brings the same documentation specificity and real-time code validation to your own coding workflow -- so your claims are review-ready before they reach any prior auth system.