Skip to main content
Customers upload files via SFTP. Each customer gets a dedicated folder structure backed by cloud storage. When files land, they are automatically parsed and converted into entities in our data engine.

​
Infrastructure

Our SFTP system provides:
  • SFTP server — Secure file upload endpoint for customers
  • Cloud storage — Three buckets per tenant:
    • Ingestion — Files land here from SFTP
    • Processed — Successfully parsed files are archived here
    • Invalid — Files that failed validation (triage queue)
  • File processor — Parses files and writes entities to our data engine

​
Customer folder structure

Each customer gets a set of default folders on the SFTP server: 
/{customer-slug}/
├── EDI/              # 837 claims and 835 remittances
├── UCRN/             # Account-to-claim crosswalk files
├── MedicalRecords/   # Supporting medical documentation
├── Contracts/        # Payer contract documents
└── Supplemental/     # Additional files
The folder name determines how files are parsed. A file in the EDI/ folder is parsed as X12 EDI. A file in UCRN/ is parsed as a pipe-delimited crosswalk.

​
Processing pipeline

1

Upload

Customer uploads a file via SFTP. It lands in the ingestion bucket.
2

Validate

The file processor validates the path format: {customerSlug}/{fileCategory}/{filename}.
3

Route & parse

Based on the folder:
  • EDI/ — X12 parser creates Claim and Remit entities
  • UCRN/ — CSV parser creates Account entities and links to Claims
  • MedicalRecords/ — Associates documents with existing Claims
4

Persist

Entities are created or updated in our data engine via the API.
5

Archive

The file is moved to the Processed bucket (or Invalid if parsing failed).

​
What happens next

Once files are parsed into entities, our background services pick up the changes in real time to:
  • Parse raw EDI data into structured Claim and Remit entities
  • Create Account entities from UCRN crosswalks and link claims
  • Link remittance payments to their corresponding claims
  • Associate medical records with patient claims