Healthcare's $36 Billion Blind Spot: Why Fixing Claims Starts Before the Encounter Ends

Every year, American health systems lose $36 billion to revenue leakage for one key reason: the documentation chain between clinical encounter and claims submission is fundamentally broken. And yet, the industry's response has been largely reactive: audits after the fact, appeals processes, revenue recovery teams chasing money that should never have been lost.

It’s a pain point healthcare administrators know all too well: fighting denials from downstream is an inefficient and costly approach to RCM challenges. Now, with Suki’s Ambient Clinical Intelligence, the real fix happens in the exam room, in real time.

A language problem disguised as a billing problem

The disconnect begins in medical school. Clinicians are trained to document in a particular dialect — "history of diabetes," "history of hypertension" — language that carries perfectly clear meaning in a clinical context. But in the coding world, "history of" is interpreted to mean the condition no longer exists. A patient who has been managing Type 2 diabetes for fifteen years can, through nothing more than standard clinical phrasing, appear on paper to be in remission.

This isn't an edge case. It's a structural misalignment baked into the documentation workflow, and it produces a cascade of downstream problems: undercoded encounters, unsupported diagnoses, and ultimately, denied claims. What looks like a billing failure is, at its source, a language failure.

"Clinical medicine and coding are inherently disconnected,” says Suki Chief Medical Officer Kevin Wang. “Anytime you hear prior authorization, or pre-cert, or utilization management, you’re already too late in the game. The reason for the clinical need for something has already been determined, yet you're trying to prove it after the fact, instead of doing it in real-time during the visit.”

The nuances compound quickly. Distinguishing between a patient's own history and a maternal family history of colon cancer, or capturing the precise stage of chronic kidney disease, requires a level of granular specificity that most ambient documentation tools aren't designed to handle. "And in this situation, engineers might say, 'Oh, that's easy — just eliminate 'history of,'" Wang said. "But then you get all this nuance where history can mean different things depending on the context."

Watch Dr. Wang’s talk, Fix It Before It’s Lost: Why the Point of Care Is the Most Valuable Step in RCM, from the AI Zone at ViVE 2026 to see why embedding Ambient Clinical Intelligence into clinical workflows solves the missing link in mid-encounter RCM.

The prior auth trap

The healthcare industry's focus on denial management is a symptom of misplaced effort, when the opportunity lies squarely upstream, at the point of care.

• 86% of claim denials are avoidable with better documentation
• 65% of denied claims are never resubmitted
• $20B spent annually just attempting to recover denied revenue

Nowhere is the upstream-downstream tension more visible than in prior authorization. Consider a 32-year-old whose parent was diagnosed with colon cancer — under existing coverage guidelines, that family history can qualify the patient for a screening colonoscopy a decade earlier than standard age thresholds. The medical necessity is real. The coverage pathway exists. And yet, when a physician orders that colonoscopy, the claim is almost automatically denied.

Why? Because the clinical context that justifies the order — the family history, the first-degree relationship, the specific diagnosis — never made it into the documentation in a form the payer's system could read and act on. The information existed in the room. It just didn't travel downstream.

This pattern plays out across screening rates with striking frequency. Colonoscopy completion rates among eligible patients sit below 70%, despite the fact that the clinical need is, in the majority of cases, already known to the treating physician. "The provider already knows it," Wang noted. "We just have to prove they've already shown it."

Ambient clinical intelligence as infrastructure

The term "ambient" has become something of a buzzword in health technology, frequently conflated with voice capture or AI-assisted transcription. Suki's framing is much more expansive. Our Ambient Clinical Intelligence (ACI) encompasses not just audio capture but intent recognition, contextual interpretation, and — critically — the ability to make real-time API calls to coding engines, adjudication systems, and audit tools while the clinical encounter is still in progress.

The implications are significant. Rather than treating documentation, coding, and claims adjudication as sequential steps in a linear process, ACI enables them to run in parallel — compressed into the window of a single encounter. By the time a physician submits a note, the relevant coding logic has already been applied, medical necessity has already been documented, and the claim is already positioned to pass first-pass adjudication.

Suki's positioning here is deliberately non-competitive with the RCM ecosystem. The company is not attempting to become a coding company or an EHR. Its bet is that it can be the intelligence layer that makes existing RCM infrastructure, like coding vendors, value-based care networks, and payers, significantly more effective by feeding them cleaner, richer, more timely clinical data.

The payer equation

Perhaps the most commercially interesting dimension of the upstream argument is its implications for payers. The conventional payer workflow is fundamentally reactive: receive claim, apply utilization management criteria, issue remit code, deny or approve. The cost of this process — query letters, supplemental documentation requests, appeals — is substantial, and largely invisible as overhead until measured against the alternative.

Wang's argument to payers is straightforward: what portion of your post-claim adjudication spend could be eliminated if the clinical data arrived before the claim, already structured, already sufficient? Early pilots with payer partners are beginning with the lowest-ambiguity cases like screening colonoscopies, mammographies, and preventive care with clear numerator and denominator coverage criteria, precisely because these are the use cases where the ROI is most legible, and the documentation requirements are least contested.

The longer-term model, if it holds, involves Suki pushing structured clinical data directly to payers at the close of an encounter, bypassing the EHR lookup entirely. Adjudication moves forward in time. Denials shrink. Chase costs fall on both sides of the transaction.

The integration challenge

None of this is simple to execute at scale. American health systems operate across a fragmented technology landscape. To give an idea of the scope and scale of the challenge, Wang says, “One of our partners at a big 18-hospital health system told me they have more than 800 active SaaS subscriptions across their organization”. Building ambient clinical intelligence that functions coherently across that heterogeneity requires a platform that is deliberately agnostic: to EHR, to large language model, to payer system.

Suki's architecture reflects that constraint. The company operates across major EHR platforms, selects language models based on task-specific performance rather than vendor preference, and is designed to interoperate with downstream coding and billing infrastructure rather than replace it. Whether that openness proves a durable competitive advantage or simply a prerequisite for market entry remains to be seen.

What is clear is that the RCM industry is entering a consolidation moment. The proliferation of ambient tools will accelerate pressure to demonstrate measurable financial outcomes rather than workflow improvements. The upstream thesis is, at its core, a financial argument: that the most efficient place to prevent a denied claim is before the encounter closes, not after the revenue is already gone.

In a sector that loses $36 billion a year to preventable leakage, the opportunity is clear.

Watch Dr. Wang’s talk, Fix It Before It’s Lost: Why the Point of Care Is the Most Valuable Step in RCM, from the AI Zone at ViVE 2026 to see why embedding Ambient Clinical Intelligence into clinical workflows solves the missing link in mid-encounter RCM.