Read any document, in any language.
Longlichi is a Vision-Language Model that extracts structured data from receipts, invoices, dispatch notes, tags, and handwritten documents.
Why Longlichi
Multi-format parsing
Handles printed forms, handwritten notes, photos of tags, crumpled receipts — real-world documents, not clean PDFs.
Multi-script support
Devanagari, Tamil, Telugu, Arabic, Latin, and mixed-script documents. One model for all.
Structured JSON output
Returns clean, typed JSON — dates as dates, numbers as numbers, ready for your ERP or database.
Number plate recognition
Automatic NPR for logistics and access control. Any lighting, any angle.
Batch processing
Process thousands of documents per hour with consistent accuracy. No fatigue, no errors.
On-premise option
Run Longlichi on your infrastructure. Document data stays within your network.
From paper to data.
Capture
- Camera, scanner, or file upload
- Supports JPEG, PNG, PDF, TIFF
- Handles skew and poor lighting
Understand
- Layout analysis and field detection
- Handwriting and print recognition
- Multi-language OCR in one pass
Extract
- Key-value pair extraction
- Table detection and parsing
- Entity recognition (dates, amounts, IDs)
Export
- Structured JSON via API
- Direct ERP/database integration
- Confidence scores per field
Documents at machine speed.
Dispatch note processing
Automate inward receipt and dispatch note data entry. Quantities, product codes, vehicle numbers — extracted instantly.
Cloth tag reading
Read fabric care labels, composition tags, and compliance markings for regulatory verification.
Invoice processing
Extract line items, totals, tax amounts, and vendor details from invoices in any format.
Logistics NPR
Read vehicle number plates for gate entry/exit logging. Works in any weather and lighting.