
AI-Powered Remote Patient Monitoring Cuts Hospital Readmissions
- 30%
- Reduction in hospital readmissions
- 12 weeks
- From scoping to clinical deployment
Prior auth isn't a billing problem, it's a staffing problem. Clinical teams spend hours each day navigating payer portals, pulling documentation, and calling payer lines to get status updates that should arrive automatically.
We build prior authorization automation software for health systems, medical groups, and digital health platforms that need the PA process running with minimal manual intervention, from criteria extraction through submission, status tracking, and denial follow-up.
Automated PA submission against payer portals using X12 278 and portal automation where EDI isn't available
Payer-specific rules engine that applies the right criteria, documentation, and code requirements per payer
Real-time status tracking so staff stop calling payer lines for updates
EHR integration that pulls clinical criteria directly from patient records
RaftLabs builds prior authorization automation software for health systems, medical groups, and digital health companies. We develop payer-specific rules engines, real-time PA status tracking, clinical criteria matching, EHR integration for criteria extraction, denial management workflows, and X12 278 submission pipelines. All prior auth automation is built HIPAA-aware with complete audit trails. Most prior authorization automation builds deliver in 10-14 weeks at a fixed cost.
Recognition
Clinical staff spending two to four hours per day completing manual prior auth forms across payer portals, time that should go to patients?
Payer-specific requirement differences causing your team to resubmit the same request multiple times because the documentation package was wrong for that payer?
Companies we've built for


Every prior authorization request that goes through a manual workflow costs a medical group two to four staff hours, pulling the patient record, reading the payer's clinical criteria, gathering supporting documentation, navigating the portal, submitting the request, and then calling to check on status because the portal doesn't send updates. Multiply that by 50 to 150 PA requests per week and you've built a department whose entire output is data entry.
The underlying process is rule-based. Each payer publishes clinical criteria guidelines for each procedure or medication code. Each request needs a defined set of documentation, diagnosis codes, ICD-10-CM codes, CPT codes, clinical notes, lab results. Submission follows a defined format: X12 278 transaction for payers who accept EDI, or portal entry for those who don't. Status updates come back on a predictable cadence. Every step follows a rule, which means every step can be automated.
We build prior authorization automation software for health systems and medical groups who need the high-volume, rule-based portions of the PA workflow handled without clinical staff involvement, and for digital health companies that need a HIPAA-aware PA layer built into their platform.
Automated prior authorization request submission across payer portals and EDI channels. For payers that accept X12 278 transactions, requests are formatted and submitted via a clearinghouse connection to Change Healthcare, Availity, or Waystar, with the payer's companion guide rules applied during transaction generation. For payers that require portal-based submission, UiPath or Microsoft Power Automate handles the portal navigation, field population, and document upload sequence, recording each interaction in the audit log.
Patient demographics, insurance identifiers, requesting provider NPI, and servicing facility NPI are sourced from your EHR via FHIR R4 Patient and Coverage resources where available, or from HL7 ADT messages for EHR systems that predate FHIR. Procedure codes (CPT/HCPCS) and diagnosis codes (ICD-10-CM) are pulled from the order or referral record. Supporting clinical documentation is attached automatically when the file is linked to the encounter. Each submission generates a reference number and timestamp in the case record so staff can see what was submitted, when, and through which channel, without opening the payer portal.
A rules engine that encodes each payer's clinical criteria requirements and maps them to the correct documentation package for each procedure and diagnosis combination. Payer clinical policy guidelines change several times per year, the rules engine version-controls each policy set so a policy update doesn't silently break in-flight requests. When a payer updates their coverage policy for a procedure, the updated rule set is applied to new requests while historical requests retain the policy version that was in effect at submission time.
The rules engine handles the differences that cause manual rework: some payers require a peer-to-peer review letter for certain CPT codes, others accept the clinical note directly; some require step therapy documentation before approving a specialty medication, others require the prescribing physician's attestation. These aren't edge cases, they're the standard operational differences across UnitedHealthcare, Aetna, BCBS plans, Cigna, and Humana that your staff currently navigates from memory. The rules engine encodes them explicitly so the correct documentation package is assembled without staff knowing which payer requires what.
Status polling against payer portals and X12 278/279 transaction responses on a defined schedule, typically every four hours during business hours, so staff can see PA status in the system without calling the payer line. X12 279 response transactions are parsed to extract approval, denial, pend, or additional information required status updates and write them back to the PA case record. Portal-based status checks use the same RPA session used for submission, navigating to the case status screen and extracting the current status without manual staff involvement.
Status transitions trigger workflow events: approval triggers notification to the scheduling team and updates the encounter authorization record in the EHR; a pend or additional information request triggers a task assigned to the clinical team with the specific information the payer is requesting; a denial triggers the denial management workflow. Time-sensitive requests, same-day procedures, inpatient admissions, get a separate escalation path with more frequent polling and immediate notification to the clinical lead when authorization isn't received within the expected window.
Automated matching of the patient's clinical record against the payer's clinical criteria for the requested procedure or medication before the PA request is submitted. For EHRs that expose FHIR R4 APIs, the matching engine reads Condition (ICD-10-CM diagnoses), Observation (lab results and vitals), MedicationRequest (current and prior medications), and DocumentReference (clinical notes) resources and checks them against the payer's criteria set. This catches missing documentation before submission rather than after denial.
For CRD (Coverage Requirements Discovery) and DTR (Documentation Templates and Rules) implementations per the Da Vinci implementation guides, supported by Epic and Cerner environments on recent versions, criteria matching is embedded in the clinician's ordering workflow via CDS Hooks. The system surfaces the payer's requirements at the point of order entry, allowing the ordering provider to confirm documentation is in place before the order is placed. For EHR environments that don't yet support CRD/DTR, the standalone criteria matcher runs as part of the pre-submission workflow rather than at order entry.
When a PA request is denied, the denial reason code from the X12 279 response or portal screen is mapped to a plain-language explanation and the standard appeal pathway for that payer and denial type. Common denial reasons, missing clinical documentation, criteria not met, incorrect procedure code, step therapy not documented, each have a defined remediation workflow in the system so the person handling the denial knows what's needed without interpreting the payer's denial code.
Appeal letter generation pulls from payer-specific templates pre-populated with patient data, clinical documentation, and the denial reason so the appeal writer starts from a complete draft rather than a blank document. Peer-to-peer review scheduling, required by many commercial payers for complex medical necessity denials, is tracked in the system with the payer's scheduling deadline and the case assigned to the requesting physician. Follow-up task queues track each appeal against the payer's appeal filing deadline so nothing ages past the timeframe. Denial trend reporting by payer, denial code, procedure code, and service line gives the clinical team the data to identify whether a specific payer is consistently applying criteria differently from their published policy.
Bidirectional EHR integration that pulls clinical data into the PA workflow and writes authorization results back to the patient record. FHIR R4 integration handles Epic (via App Orchard), Cerner Millennium, and Athenahealth, Patient, Coverage, Condition, Observation, MedicationRequest, DocumentReference, and ServiceRequest resources cover the clinical and insurance data the PA workflow needs. SMART on FHIR OAuth 2.0 handles authorization for FHIR-enabled environments.
For EHR systems running on HL7 v2 interfaces, the integration engine (Mirth Connect or Azure Health Data Services) parses ADT, ORM, and ORU message types to extract the equivalent clinical data. Authorization results, approval number, approved dates, approved units, payer reference number, are written back to the EHR as FHIR R4 Coverage resources or via HL7 interface depending on the EHR's write API support. This means the authorization number lives in the patient's EHR record and flows automatically to the claim at billing rather than being transcribed manually by the billing team.
Prior authorization automation software handles the PA workflow, identifying which procedures need authorization, gathering clinical criteria, submitting requests to payers, tracking status, and managing denials, with minimal manual staff involvement. Off-the-shelf PA software products exist, but they're built around average payer behavior. A custom build makes sense when your payer mix has specific portal requirements that generic tools don't cover, when your EHR is the source of clinical criteria that need to feed the PA workflow automatically, when you operate a digital health platform that needs a PA layer as a service component rather than a standalone product, or when your current denial rate is high enough that the revenue and staff cost justify building the right system.
Each payer's clinical criteria, documentation requirements, and submission preferences are encoded in the rules engine separately. The rules engine holds a policy version per payer per procedure code set, when a payer updates their clinical policy, the update is applied to new requests while historical requests retain the policy version in effect at the time of submission. The engine handles the specific differences that cause manual rework: which payers require step therapy documentation before specialty medication approval, which payers require peer-to-peer review for specific CPT codes, which payers need the prescribing physician's attestation rather than the clinical note, and which payers accept X12 278 EDI versus portal-only submission. The payer list and requirement details you give us in the scoping session determine what gets encoded in the initial build.
All prior authorization automation we build is designed to meet HIPAA technical safeguard requirements under 45 CFR Part 164. PHI that flows through the automation, patient demographics, diagnosis codes, clinical documentation, insurance identifiers, is handled with least-privilege access controls: each process component is scoped to the minimum data access it needs for its specific function. Credentials for payer portal sessions and EHR API connections are stored in a secrets manager (AWS Secrets Manager or HashiCorp Vault), not in configuration files or environment variables.
Every automated action generates a structured audit log entry: what data was accessed, what was submitted, to which payer, at what timestamp, and with what outcome. Audit logs are stored in append-only tamper-evident storage (CloudTrail on AWS) with 6+ year retention. Data in transit between the automation system and payer portals or clearinghouses is encrypted in transit using TLS 1.2 minimum. PHI at rest is encrypted using AES-256. Business Associate Agreements with all infrastructure providers that process PHI are executed before any PHI enters the system. We deliver the data flow documentation as part of the project so your compliance team can review the automation against your existing HIPAA policies.
Most prior authorization automation builds deliver in 10-14 weeks at a fixed cost. The timeline depends on three variables: the number of payers in scope, the complexity of the EHR integration, and whether the clinical criteria matching needs to connect into the ordering workflow at the point of care or can run as a pre-submission step. A build scoped to three to five payers with a single FHIR R4 EHR integration is at the 10-week end. Adding more payers, portal automation for payers without EDI, or CRD/DTR integration at the EHR order-entry level extends the timeline. We confirm the specific scope and fixed cost during the discovery session before committing to a delivery date.
Yes. Authorization results, approval number, approved dates, approved procedure codes, approved units, and payer reference number, are written back to the patient's EHR record via FHIR R4 Coverage or Claim resources for EHRs that support FHIR write operations, or via HL7 v2 interface for older systems. This means the authorization number is in the EHR record when the claim is generated, and the billing team doesn't need to manually look up or transcribe it from the payer portal. For Epic environments, the authorization data can be written back into Epic's coverage record so it flows through to the claim in Epic's billing workflow. The write-back approach for your specific EHR version and configuration is determined during the integration scoping phase.
What clients say
Three-year average engagement. Founders and operators describing the work in their own words. No marketing varnish.

All of the sprints were completed on schedule and on budget. We highly recommend RaftLabs!
Tell us your payer mix, your EHR, and where the PA workflow breaks down. We'll scope a system and give you a fixed cost.