Documents in, structured data out.
The engine behind “add a shipment by AI” and declaration pre‑fill in a customs platform — reusable for invoices, delivery notes and any form-heavy paperwork.
Context
Customs and freight paperwork arrives as PDFs, scans and spreadsheets in three languages. Invoices, bills of lading and packing lists have to be keyed into declarations and shipment records before anything else can happen.
Challenge
Every supplier lays out its documents differently, so a fixed OCR template breaks on the next shipment. The same engine has to serve a cloud-hosted platform and customers whose documents may not leave their own premises — and every extraction has to be cached, metered and attributable to a tenant.
What we built
- A template-driven extraction service: submit a file or text, receive structured JSON
- Async jobs, result caching by file hash and per-tenant usage metering
- Admin UI for defining and versioning extraction templates
- A document conversion pipeline for PDF, images and Excel
- Model back-ends: Gemini, OpenAI, or locally hosted Qwen vision models for on-premises deployments
- Two implementations of the same contract: PHP and .NET
Outcome
The engine behind "add a shipment by AI" and declaration pre-fill in the customs platform; reusable for invoices, delivery notes and any form-heavy paperwork.
Architecture
Tell us what your business needs.
Speak directly with our solution architects and get a practical architecture, estimate and delivery plan.