Processing Commercial Invoices

Overview

CargoLint’s invoice processing engine automatically extracts structured data from commercial invoices with high accuracy. Our AI-powered system handles various invoice formats and provides confidence scores for every extracted field.

Supported Invoice Formats

CargoLint accepts invoices in the following file formats:

• PDF - Native PDF documents
• TIFF - Multi-page TIFF files
• JPEG/PNG - Single-page images

Maximum file size: 5 MB per document.

Extracted Fields

When processing an invoice, CargoLint extracts the following key information:

Core Information

• Shipper/Seller - Company name, address, contact details
• Consignee/Buyer - Company name, address, contact details
• Invoice Number - Unique invoice identifier
• Invoice Date - Issue date and terms
• Currency - Transaction currency code
• Total Amount - Invoice total with tax breakdown

Line Items

Each line item includes:

• Item description
• Quantity and unit of measure
• Unit price and line total
• HS code classification
• Country of origin
• Weight and dimensions (if available)

Confidence Scoring

Each extracted field includes a confidence score (0.0 - 1.0) indicating extraction reliability:

{
  "invoice_number": {
    "value": "INV-2026-001234",
    "confidence": 0.98
  },
  "shipper": {
    "name": "TechExport Co., Ltd",
    "confidence": 0.95,
    "address": "Shanghai, China",
    "address_confidence": 0.92
  },
  "line_items": [
    {
      "description": "Electronic components",
      "quantity": 1000,
      "unit_price": 25.50,
      "hs_code": "8542.31.00",
      "hs_code_confidence": 0.87
    }
  ]
}

Invoice Confidence Thresholds

• High confidence: > 0.85
• Review recommended: 0.65 - 0.85
• Low confidence: < 0.65 (requires manual review)

Compliance Checks

CargoLint automatically validates:

• Required field presence
• HS code consistency with product description
• Total amount calculations
• Incoterm validation
• Weight mismatches
• Export restrictions by destination country

Handling Low-Confidence Extractions

When the overall document confidence falls below 70%:

1. Review Queue - Documents are automatically flagged for human review
2. Manual Correction - Team members can review and correct extractions in a side-by-side viewer
3. Learning - Corrections feed back into the system to improve future accuracy

Best Practices

• Ensure invoice images are clear and well-lit
• Avoid blurred or skewed documents
• Provide complete invoices (don’t crop important fields)
• Use standard invoice templates when possible
• Scan at 300 DPI minimum for best results

Processing API Endpoint

Upload an invoice for processing:

curl -X POST https://api.cargolint.com/api/v1/documents \
  -H "X-API-Key: YOUR_API_KEY" \
  -F "file=@invoice.pdf"

The API returns the document ID and processing status. Retrieve extraction results once processing completes:

curl -X GET https://api.cargolint.com/api/v1/documents/{document_id}/extractions \
  -H "X-API-Key: YOUR_API_KEY"

Processing typically completes within 2-5 seconds for standard invoices.