Revedy Logo
Reference Document

The Revedy Workflow: Architecture & Lifecycle Overview

The Revedy Revenue AI Platform operates as an integrated, five-stage pipeline that governs the medical coding and billing lifecycle from initial clinical documentation receipt to post-submission payer response and appeals. This reference document defines each stage of the workflow, the underlying clinical and regulatory validation engines, and the human-in-the-loop governance controls that ensure audit-ready revenue integrity.

Looking to evaluate this workflow with your organization's clinical data? Contact us for immediate demo access or sandbox API credentials.

1.End-to-End Pipeline Architecture

Traditional revenue cycle management (RCM) relies on disconnected point solutions for computer-assisted coding, claim scrubbing, coverage verification, and denial management. When these functions operate in silos, clinical evidence captured during encounter documentation is lost before claim adjudication—resulting in preventable medical necessity rejections, NCCI bundling errors, and high administrative costs to collect.

Revedy unifies proactive pre-submission claim scrubbing and reactive post-submission appeals automation against a single Healthcare Knowledge Graph. Every stage preserves verbatim clinical provenance, linking procedures, diagnoses, modifiers, and payer policies into a single auditable record.

Figure 1.0 — Revedy Core Coding, Medical Necessity & Payer Response Pipeline

Loading diagram...

2.Workflow Stage Specifications

Each encounter progresses sequentially through five distinct processing stages. Below is the operational specification for each stage illustrated in Figure 1.0, detailing what the engine ingests, how rules and policies are evaluated, and what artifacts are produced.

Stage

01

Multi-channel document ingestion, classification, and clinical parsing.

Receive Case Documentation

The workflow begins when clinical and administrative case files are transmitted to Revedy via our RESTful API (V2/V3 endpoints), automated SFTP batch pipelines, or the web-based case portal. The platform normalizes heterogeneous file formats—including unstructured EHR progress notes, multi-page operative reports, neurophysiological monitoring logs, lab panels, and scanned PDFs—into a structured encounter context.

Ingests
Accepts clinical notes, operative reports, diagnostic logs, CMS-1500 drafts, and audio transcripts via API, SFTP, or direct upload.
Executes
Executes structural document parsing (tables, headers, timestamps, signatures) with automated Optical Character Recognition (OCR) fallback for scanned PDFs and faxes.
Executes
Classifies each artifact by document type (ClinicalDocument, OperativeNote, DiagnosticReport, PayerResponse) and identifies the clinical specialty for downstream model routing.
Outputs
Produces a unified, indexed encounter record with character-level text offsets preserved for citation highlighting.

Stage

02

Procedural extraction, client rule customization, and carrier edit validation.

Code CPT (Procedural Coding & Rule Engine)

Once the encounter context is assembled, Revedy's agentic CPT engine executes a three-tier procedural analysis. Rather than relying on static keyword matching, an orchestrator agent selects specialized clinical tools—such as E/M medical decision-making (MDM) calculators, time-unit aggregators for IONM monitoring, or multi-vessel surgical parsers—and then layers organization-specific and payer-specific rules before finalizing the procedural code set.

Step 2.1
Identify CPT codes, units, and modifiers: Extracts primary and add-on CPT/HCPCS codes, calculates time-based or anatomical units, and assigns appropriate modifiers (e.g., -25, -26, -59, -TC, -95) backed by verbatim quotes from the clinical record.
Step 2.2
Apply customer-specific rules: Enforces organization-defined coding protocols, specialty templates, internal compliance thresholds, and claim-splitting rules (such as separating technical and professional components).
Step 2.3
Apply payer-specific rules: Evaluates the candidate code set against Medicare National Correct Coding Initiative (NCCI) Procedure-to-Procedure (PTP) edits, Medically Unlikely Edits (MUEs), and carrier-specific code substitutions (for example, converting CPT 95941 to HCPCS G0453 for Medicare claims).
Outputs
Outputs a claim-aware procedural code set with RVU calculations, modifier justifications, and line-item audit trails.

Specialty & Capability Reference

Proactive Claim Scrubbing & Undercoding Detection

During CPT extraction, the engine simultaneously audits for both overcoding that triggers bundling denials and undercoding (missed add-on codes or unbilled time units) that causes revenue leakage. See Denial Management & Claim Scrubbing for detailed rule coverage.

Stage

03

ICD-10-CM diagnosis extraction, specificity resolution, and pointer mapping.

Code ICD-10 (Diagnosis Assignment & Specificity)

In Stage 03, the engine assigns ICD-10-CM diagnosis codes in strict accordance with the official CMS and CDC Coding Guidelines required under HIPAA. The engine resolves documented conditions to the highest level of clinical specificity (laterality, acuity, episode of care, and manifestation) and validates relational constraints across all diagnoses on the claim.

Executes
Extracts diagnoses, signs, symptoms, and pertinent history codes supported by the provider's clinical documentation for the Date of Service (DOS).
Validates
Enforces ICD-10-CM structural conventions, including Excludes1 (mutually exclusive conditions that must never be billed together), Excludes2, and Code First / Use Additional sequencing rules.
Validates
Links CMS-1500 Box 21 diagnosis codes to individual service lines in Box 24E via diagnosis pointers to establish line-level clinical justification.
Surfaces
Supports three configurable operational modes—Literal Only (Mode 01), Standard Outpatient Guidelines (Mode 02), and Candidate Diagnoses with AHIMA/ACDIS Compliant Queries (Mode 03).

Companion Reference Document

Configurable ICD-10-CM Coding Modes (Modes 01, 02, and 03)

Organizations can configure how conservatively the engine assigns diagnosis codes and generates physician queries. Review the companion specification, ICD-10-CM Coding Modes Reference Document, for worked examples and regulatory citations.

Stage

04

Pre-submission policy cross-referencing, gap detection, and risk scoring.

Medical Necessity Validation & Risk Scoring

Before a claim is released for billing, Stage 04 evaluates the combined CPT/HCPCS and ICD-10-CM code pairings against Revedy's Healthcare Knowledge Graph. This graph continuously indexes CMS Local Coverage Determinations (LCDs), National Coverage Determinations (NCDs), Medicare Coverage Articles, and commercial payer medical policies by state and carrier.

Step 4.1
Use payer guidelines: Retrieves the exact LCD, NCD, or commercial medical policy governing the procedure for the patient's payer, jurisdiction, and date of service.
Step 4.2
Identify gaps: Flags missing clinical prerequisites (such as conservative therapy duration, prior imaging requirements, or non-covered diagnosis pairings) that would cause a first-pass medical necessity denial.
Step 4.3
Identify potential diagnosis: Scans the full encounter record for documented clinical findings or secondary diagnoses that legitimately satisfy payer coverage criteria—or initiates a compliant provider query when clarification is required.
Step 4.4
Provide overall risk scores: Calculates a quantitative pre-submission approval confidence and risk score, routing high-confidence claims to billing queues and flagging edge cases for human-in-the-loop review.

Capability Reference

Automated Medical Necessity Verification

Explore how our Healthcare Knowledge Graph maps procedures to covered diagnosis lists across Medicare and commercial plans in Automated Medical Necessity Validation and Medical Necessity for Medical Coding.

Stage

05

Automated denial ingestion, policy synthesis, and appeal generation.

Payer Response & Automated Appeals

When a payer returns an adjudication response, Explanation of Benefits (EOB), or claim denial notice, Stage 05 closes the revenue cycle loop. Automated SFTP and API listeners match incoming payer correspondence to the original encounter, evaluate the payer's denial rationale against clinical evidence, and assemble an audit-ready appeal package.

Step 5.1
Process payer response: Ingests 835 Electronic Remittance Advice (ERA) files, EOB PDFs, and denial letters, extracting Claim Adjustment Reason Codes (CARCs), Remittance Advice Remark Codes (RARCs), and free-text payer rationales.
Step 5.2
Review guidelines, guidance, and materials: Cross-references the payer's stated denial reason against the original clinical documentation, coding rules (NCCI/MUE), and the governing LCD, NCD, or commercial policy in effect on the date of service.
Step 5.3
Generate response, appeal, etc.: Drafts an evidence-based appeal letter or corrected claim recommendation that directly quotes the patient's medical record and cites the exact section of the payer's own coverage policy.

Capability Reference

Turning Denials into Defensible Revenue Recovery

Learn how automated evidence synthesis replaces generic appeal templates in Payer Response & Appeals Automation.

3.Data Artifacts & Knowledge Sources by Stage

The table below summarizes the primary inputs, authoritative knowledge graph datasets, and structured outputs generated across each stage of the Revedy workflow.

StagePrimary InputsAuthoritative Sources QueriedOutput Artifacts & Gates

01

Ingestion

EHR clinical notes, operative reports, diagnostic logs, DOCX/PDF files, audio transcriptsDocument classification taxonomy; specialty-detection models; OCR extraction pipelines
Indexed Encounter JSONSource Span Map

02

Code CPT

Indexed encounter text, timestamps, anatomical sites, payer/carrier metadataAMA CPT®, CMS HCPCS, CMS RVU & CLFS fee schedules, NCCI PTP edits, MUE tables, client rulebook
CPT / HCPCS LinesUnits & ModifiersNCCI Audit Log

03

Code ICD-10

Assessment/plan, problem list, signs/symptoms, abnormal lab and imaging findingsCDC/CMS ICD-10-CM Official Guidelines, Excludes1/Excludes2 tables, AHIMA/ACDIS query standards
Box 21 ICD-10-CM SetBox 24E PointersGATE:Provider Query (Mode 03)

04

Medical Necessity

Proposed CPT/ICD-10 pairings, clinical evidence spans, payer & state jurisdictionCMS Medicare Coverage Database (LCDs, NCDs, Articles), commercial medical policies
Claim Risk ScoreCoverage Gap ReportGATE:ICAT Auditor Queue

05

Payer Response

835 ERA files, EOBs, CARC/RARC denial codes, payer denial lettersOriginal encounter record, DOS-specific LCD/NCD/commercial policy, NCCI & CPT guidelines
Denial Root-Cause BriefCitation-Backed Appeal Letter

4.Governance, Human-in-the-Loop & Security Controls

Across all five stages of the workflow, Revedy enforces strict architectural safeguards to prevent “black-box” AI behavior, protect against False Claims Act (FCA) exposure, and maintain enterprise-grade data security.

Human-in-the-Loop (ICAT) Governance

Revedy is built to augment—never blindly replace—certified coders and clinical auditors. Through our Human-in-the-Loop (HITL) Intelligent Coding Audit Tool (ICAT), low-confidence encounters or policy exceptions are automatically routed to review queues. Auditors inspect AI recommendations side-by-side with highlighted clinical evidence and can approve or refine codes in a single click. Organizations can also augment their coding workforce with dedicated AI coding agents to scale throughput without sacrificing human oversight.

Verbatim Clinical Provenance

Every CPT code, unit calculation, modifier, ICD-10-CM diagnosis, and medical necessity determination generated by the workflow is permanently linked to its exact character span in the source documentation, the governing rule or policy citation, and a cryptographic timestamp. No code is assigned without traceable clinical evidence.

Temporal Policy & Code Versioning

Healthcare coding sets and payer coverage rules change quarterly and annually. The Revedy Healthcare Knowledge Graph tracks chronological valid_from and valid_to dates for all ICD-10-CM updates, NCCI edits, and Medicare/commercial policies—ensuring both pre-submission scrubbing and retrospective appeals are evaluated against the rules in effect on the Date of Service.

HIPAA Compliance & Enterprise Security

All workflow stages execute within a strictly HIPAA-compliant security architecture backed by Business Associate Agreements (BAAs), HITRUST CSF readiness, OAuth 2.0 authentication, and end-to-end encryption in transit and at rest. For health systems requiring complete data residency control, the entire workflow can also be provisioned via dedicated on-premises or private cloud deployment.

5.Related Reference Documentation & Capabilities

For deeper technical specifications on individual components of the Revedy workflow, consult the following platform documentation and capability guides.

(1)

ICD-10-CM Coding Modes Reference Document

Detailed regulatory specification of Revedy's three configurable ICD-10-CM coding modes (Mode 01 Literal Only, Mode 02 Standard Outpatient Guidelines, and Mode 03 Candidate Diagnoses with AHIMA/ACDIS Compliant Queries), including worked clinical examples.

https://revedy.io/icd10-coding-modes

(2)

Human-in-the-Loop (HITL) Governance & Specialty AI Insights

Architectural overview of the Intelligent Coding Audit Tool (ICAT), side-by-side citation verification, auditor approval workflows, and specialty-specific denial prevention benchmarks.

https://revedy.io/ai-insights

(3)

Core Platform Capability Guides: Coding, Medical Necessity & Payer Response

Deep-dive functional overviews for each core stage of the revenue cycle pipeline: Pre-Submission Claim Scrubbing & Denial Prevention, Automated Medical Necessity Validation, and AI Payer Response & Appeals Automation.

https://revedy.io/capabilities/coding

(4)

Security, HIPAA Compliance & API Integration Documentation

Enterprise security specifications covering HIPAA compliance, HITRUST CSF readiness, and SOC 2 controls, data security protocols, and RESTful API integration endpoints.

https://revedy.io/security

© 2026 Revedy, Inc.

Privacy Policy