What is Invoice OCR?
Invoice OCR combines AI vision models with sophisticated validation logic to convert invoice documents into structured, machine-readable data. The platform processes both images and PDFs, extracting everything from line items to tax calculations while ensuring mathematical accuracy.Invoice OCR uses OpenRouter to access multiple state-of-the-art vision models, giving you flexibility to choose the best model for your use case.
Key Capabilities
Multiple Extraction Modes
Choose the right extraction mode for your workflow:- Raw Text OCR - Extract unstructured text from documents for basic processing
- Structured JSON - Convert invoices into the MyBillBook schema format
- India GST (v4) - Specialized extraction with automatic reconciliation for Indian GST invoices
Intelligent Reconciliation
The reconciliation engine validates extracted data against invoice totals:- Automatically verifies line item calculations (quantity × rate)
- Validates discount applications (sequential and flat discounts)
- Confirms GST calculations (CGST/SGST for intra-state, IGST for inter-state)
- Detects and applies round-off adjustments
- Flags mismatches exceeding ₹0.05 tolerance
PDF and Image Support
Process documents in multiple formats:- Images: PNG, JPG, JPEG via base64 encoding
- PDFs: Via public URL or base64 data URL
- Multi-page PDF support with cross-page tax table validation
India GST Compliance
Purpose-built for Indian invoicing:- Recognizes standard GST slabs (0%, 0.25%, 3%, 5%, 12%, 18%, 28%)
- Validates GSTIN format and place of supply
- Handles HSN/SAC codes and UQC units
- Processes header-level and line-level discounts
- Supports TCS (Tax Collected at Source)
- Manages additional charges (freight, insurance, etc.)
How It Works
Upload Document
Upload an invoice image or PDF through the web interface or API. The system accepts files up to 10MB.
AI Processing
Your document is sent to an OpenRouter vision model (default: Gemini 2.5 Flash) with a specialized prompt that enforces strict JSON schema compliance.
Reconciliation
The reconciliation engine validates all calculations, applies discount sequences, splits GST amounts, and ensures totals match printed values within ₹0.05 tolerance.
API Architecture
Invoice OCR provides three REST API endpoints:| Endpoint | Purpose | Schema |
|---|---|---|
/api/ocr | Raw text extraction | Plain text response |
/api/ocr-structured | MyBillBook schema | Compact JSON format |
/api/ocr-structured-v4 | India GST normalized | v4 schema with reconciliation |
Technology Stack
Invoice OCR is built with modern web technologies:- Framework: Next.js 15.5.2 with App Router
- Language: TypeScript 5 (strict mode)
- UI: React 19, Tailwind CSS v4, shadcn/ui
- OCR Provider: OpenRouter API with vision models
- Testing: Vitest with comprehensive reconciliation tests
Use Cases
Accounts Payable Automation
Automate invoice processing workflows:- Extract vendor details, line items, and amounts
- Validate tax calculations before payment
- Route invoices based on extracted metadata
Expense Management
Digitize receipts and invoices for expense tracking:- Capture itemized expenses from scanned documents
- Validate GST claims with automatic reconciliation
- Export to accounting systems in structured format
Tax Compliance
Ensure GST compliance and audit readiness:- Verify GSTIN and HSN codes
- Validate tax calculations against GST slabs
- Generate audit trails with reconciliation reports
Data Migration
Digitize legacy invoice archives:- Batch process historical invoices
- Extract structured data for database migration
- Maintain data quality with validation checks
Next Steps
Quick Start
Process your first invoice in under 5 minutes
Installation
Set up Invoice OCR in your development environment