Skip to primary interaction
Healthcare

Remit / Denial Intake

Catch remit discrepancies during intake, not after posting

  • Extracts payment, payer, provider, and claim fields with confidence signals
  • Routes low-confidence fields and discrepancies to review before posting
  • Turns remit intake into a real-time reconciliation queue
Review selected remitSee reconciliation queue

Primary interaction

Default path

Selected remit first

Show extracted fields, confidence, and reconciliation flags before the user scans the full corpus.

Document entersFields extractConfidence routesRecon issues surface
Select a document above to see the extraction side-by-side.
Failed to load documents.

Posting issues usually surface after the damage is done. This queue catches them during intake.

Loading reconciliation queue…
Under the hood

Capability proof

Capability proof

Document intake and reconciliation workflow

Service model

Live document-intake workflow for remits, extraction detail, and reconciliation queues.

Intelligence layer

Extracts payment fields, scores confidence, and surfaces discrepancies.

Operational state

Maintains document-level extraction results and reconciliation exceptions.

Human control

Low-confidence fields and payment mismatches remain visible for review.

Business value

Catches posting issues during intake instead of after cash damage compounds.

Why a real-time queue compounds

Remit / Denial Intake — Live AI Demo

Remit intake creates posting lag in nearly every health system's RCM stack. Manual read-and-key is slow, but the larger issue is that reconciliation problems often stay hidden until cash posting. The win is cycle-time compression: extracted fields, confidence signals, and discrepancies become visible while the work can still be routed.

Economics

Loading economics…

Architecture

Estimated monthly cost at portfolio traffic: under $1/month (Lambda free tier + minimal API Gateway requests)

LayerTechPurpose
Synthetic dataJSON · 50 docs (40 ERA + 10 EOB) · deterministic seedReal 835 segment shape, paper-EOB OCR artifacts, confidence-scored fields
ComputeAWS Lambda (nodejs20.x · 512MB)Single function, route-based dispatch across 3 endpoints
APIAPI Gateway (HTTP API)Shared SAM stack with other Workshop demos; path-parameter routing for document fetch
FrontendReact + CSS modules · EDGE design tokensSide-by-side raw / structured view, reconciliation queue, confidence flagging
DeployAWS SAM · us-east-2`sam build && sam deploy`; CloudFormation manages all resources

What this demo is, and isn't

What this demo is, and isn't

  • All remit data is synthetic. NPIs use a prefix (9) that is not in any real NPI registry range. Claim numbers, payer trace numbers, and check numbers are generated from a deterministic seed.
  • The 835 segment shapes (ISA / GS / ST / BPR / TRN / CLP / CAS / SVC) and CARC adjustment reason codes (CO-45, PR-1, CO-97, etc.) are real WPC ANSI X12 values, included so the response shape is accurate. Real production data never appears in this demo.
  • The EOB OCR artifacts (substituted characters, comma-for-period on amounts) and confidence score distributions are simulated to reflect real document AI behavior. In production, confidence varies by payer, font quality, and document source.
  • The economics math assumes a fully-automated extraction workflow. Real implementations typically reach 70–90% automation, with the remainder requiring human review (which is what the confidence flagging surfaces in this demo).