AI FSA HSA Claim Processing Agent

How Cassidy automates this using AI

Step 1: Trigger on claim submission or card transaction

The Workflow activates when a new claim arrives—whether uploaded through a member portal, received via email, ingested from WEX/Alegeus card transaction feeds, or pulled from carrier EOB files.

Step 2: Extract and normalize claim data

Cassidy uses OCR to extract key fields from receipts and EOBs: merchant name, date of service, itemized expenses, CPT codes, patient details, and payment amounts. Data is normalized and matched to the participant, dependent, plan year, and account type (medical FSA, LPFSA, DCFSA, or HSA).

Step 3: Screen for 213(d) eligibility and auto-substantiation

Cassidy checks each expense against IRS 213(d) rules and applies auto-substantiation methods: IIAS/EPL validation for retail purchases, copay matching against plan design, recurring expense recognition, and EOB verification. Claims that pass are flagged for auto-approval.

Step 4: Apply plan-specific rules and adjudicate

The Workflow runs claims through your plan's rules engine—enforcing grace periods, carryover limits, run-out windows, LPFSA/HSA ordering, dependent care service periods, and annual caps. Claims are auto-approved, pended for documentation (receipts, LMN), or denied with standardized reason codes.

Step 5: Execute reimbursement and sync to benefits platforms

Approved claims trigger ACH or check payments to participants. Cassidy generates funding files and remittance advices, syncs adjudication statuses back to WEX or Alegeus, and initiates repayment workflows for unsubstantiated card transactions.

Step 6: Log audit trail and generate compliance artifacts

Every decision—approval, pend, or denial—is logged with supporting documentation, extracted data, and reason codes. Cassidy maintains HIPAA-compliant retention for IRS and ERISA audits, and produces employer reporting on auto-adjudication rates, pend aging, and denial trends.