On June 18, 2026, R1 — the healthcare revenue management company that works with 95 of the top 100 U.S. health systems — announced two new capabilities inside its Phare OS platform: Payer Atlas and Phare Intelligence. The stated goal is to resolve claims “as care happens, not weeks later.” That framing deserves attention, because it implies a fundamental shift in how medical coding fits into the revenue cycle.
Why Claims Resolution Takes So Long
The typical claim lifecycle unfolds in a sequence that hasn’t changed much in decades. A patient encounter is documented, a coder assigns diagnosis and procedure codes, the claim is constructed and submitted, a payer reviews it against a policy rulebook, and — weeks later — a payment or denial arrives. At each handoff, information is translated from one language to another: clinical to coding, coding to claim, claim to payer interpretation.
That translation chain is where friction accumulates and denials are born. Payers and providers have historically operated from different data models, and every mismatch in how a clinical fact is coded versus how a payer expects to see it documented creates the conditions for a rejection. Medical coders sit at one of the most consequential translation points in that chain.
Payer Atlas: 1,500 Payer Connections at Scale
The first new Phare OS capability is Payer Atlas, a proprietary intelligence layer with more than 1,500 payer connections and more than 600 million payer transactions processed annually. The premise is that payer behavior — what gets denied, under what circumstances, and by which plans — is learnable when you have enough transactional data. Payer Atlas converts that accumulated intelligence into prospective guidance that providers can act on before a claim is submitted.
In practice, this means code combinations can be pre-validated against known payer behaviors at the point of coding. For medical coders, that creates a qualitatively different kind of feedback loop. Instead of learning that a claim was denied three weeks ago, the system can flag in real time that a particular code pair tends to trigger a medical-necessity denial with a specific payer in a specific service line. The correction happens before the claim leaves the building.
Phare Intelligence: Reading the Entire Medical Record
The second new capability, Phare Intelligence, addresses the most persistent limitation of AI-assisted coding tools: their reliance on structured data fields and keyword matching rather than the full clinical narrative. Phare Intelligence reads the entire medical record — unstructured notes, operative reports, discharge summaries, pathology findings — and interprets the record holistically to produce:
- Accurate medical necessity determinations linked to clinical evidence in the record
- ICD-10-CM diagnosis codes and procedure codes derived from full clinical context
- Appeal justifications grounded in specific documentation from the patient’s record
- Flags for documentation gaps that could affect adjudication before the claim is submitted
The distinction matters because clinical documentation rarely maps neatly to code descriptions. A surgical note might describe a complication in plain language without using ICD-10 terminology. A discharge summary might reference a comorbidity that changes DRG assignment, but only if the reader processes the entire document rather than scanning for keywords. Phare Intelligence is designed to catch those cases — the ones where coding accuracy depends on reading what a clinician actually wrote, not just matching structured fields.
As Dr. Martin Seneviratne, Co-CEO of R37 (R1’s AI innovation lab), put it in the announcement: “The revenue cycle has been stuck in a reactive, transactional model for decades, with providers and payers locked into an expensive back-and-forth that serves neither.”
What Real-Time Adjudication Means for Medical Coders
The long-term vision behind Phare OS — real-time adjudication — would compress the claim lifecycle dramatically. Rather than coding completing an encounter, a claim being submitted, and corrections happening weeks later in response to denials, the system would align clinical documentation, coding, and payer policy in or near real time. A claim would be effectively adjudicated before it reaches the payer, because every factor that influences payer decisions has already been addressed.
For medical coders, that shift creates both new constraints and new value. The window for catching errors narrows. The expectation shifts from fixing denials reactively to preventing them prospectively. Coders who understand how payer logic maps to coding decisions — not just how to assign codes — will be the ones best positioned to add value in that environment. Coding review becomes a real-time quality control function rather than a pre-billing audit task.
The Scale That Makes Payer Intelligence Actionable
Phare OS is currently live across R1 customer organizations representing more than $76 billion in Net Patient Revenue. That scope gives R1 something most point solutions cannot offer: a transaction dataset large enough to make payer behavior genuinely predictable at the code level. With 600 million payer transactions flowing through Payer Atlas each year, patterns that would be invisible to a single health system — a payer’s tendency to deny a specific E&M level for a particular diagnosis, for instance — become detectable and actionable intelligence across the platform.
This is the infrastructure play behind real-time adjudication. Coding accuracy at scale isn’t just about knowing the guidelines; it requires knowing how specific payers respond to specific code combinations in specific service lines. That knowledge requires transaction data at a volume that individual health systems rarely accumulate on their own.
What CDI Teams Should Watch
If Phare Intelligence reads entire records to generate codes and medical necessity determinations, the completeness and clarity of clinical documentation becomes more consequential than ever. Vague language in an operative note doesn’t just create coding risk — it creates adjudication risk at the moment a claim enters the system. CDI teams that have focused on documenting principal diagnosis and CCs for DRG optimization will need to expand their frame to include the full clinical narrative that AI tools now consume directly.
The shift underway is from documentation as a billing prerequisite to documentation as the primary data input for an AI-driven adjudication system. Getting documentation right the first time is no longer just about compliance — it is the mechanism by which real-time adjudication becomes possible.
Preparing for the Next Phase of Revenue Cycle
The R1 Phare OS update signals where the broader market is heading, even for organizations that aren’t R1 customers. Payer-specific coding intelligence is becoming a baseline expectation, not a differentiator. Full-record reading is supplanting keyword matching as the standard for AI-assisted coding. And real-time adjudication is the benchmark against which revenue cycle performance will increasingly be measured.
Medical coders who understand both the clinical context and the payer landscape — and who can work alongside AI tools rather than around them — are best positioned for that transition. Medikode’s automated medical coding platform is built for exactly this shift, combining AI-assisted coding with real-time accuracy feedback so providers are prepared as the industry moves toward real-time adjudication.