Automating Import and Export Trade Declarations

A structured document-processing workflow that reduces manual re-entry and identifies validation issues before declaration data reaches the existing filing process.

Authorised staff remain responsible for reviewing the prepared data and completing government submission. The application supports that process; it does not guarantee regulatory acceptance.

Brightex Pvt Ltd handles imports and exports involving Hong Kong, where shipment information must be prepared for government trade declarations. Staff were collecting details from shipment PDFs and entering them into the declaration workflow manually. Documents varied in layout, and a mismatch or missing detail could require additional review before filing. Satani Ventures built an on-premise application that extracts structured fields from shipment PDFs, applies validation and calculation rules, and prepares the data for review by authorised staff. Jobs with validation errors or unresolved review flags are held back rather than treated as ready for filing. The application works alongside Brightex's existing desktop filing software. Staff review the extracted information, correct issues and use the approved output in the established filing workflow.

Client: Brightex Pvt Ltd
Place: Hong Kong
Input: Shipment PDF per job
Deployment: On-premise (office LAN)

How it works

Authorized staff sign in to the internal declaration application on the office network.
They upload one PDF per shipment and choose import or export.
A queued job reads the document and extracts structured fields from what is printed, without inventing unsupported details.
When a courier tracking number is present, public carrier tracking may fill transport gaps; if lookup fails, processing continues from the PDF alone.
Currency conversion, line values, and compliance rules run in deterministic code against local master data.
Validation errors or review flags keep the job in a needs-review state — it is not eligible to send.
Authorised staff compare the form to the original PDF, correct fields, and save; the filing interface file is regenerated from approved data.
When checks pass, staff hand the prepared file to their existing desktop filing software to complete government submission.

Services involved

Business Automation & Integrations
Replaced manual copy-paste from heterogeneous shipment PDFs with a queued pipeline that extracts, enriches, validates, and produces filing-ready output with explicit stop points when data does not comply.
Custom Software Development
Built the internal web application — job queues, review forms, validation UX, and operational tooling — around Brightex's import and export declaration workflow.
Cloud & Infrastructure
Deployed on Brightex's on-premise network so commercial PDFs and declaration data stay inside their boundary, with authenticated access and background workers.

The challenge: accurate declarations from varied documents

Brightex needs to prepare import and export declarations using information spread across invoices and shipping documents. Staff were collecting the required details, checking them against the source documents and re-entering the information into the existing filing software. The challenge was not simply to fill in a form faster. It was to keep declaration data consistent with the supporting documents while identifying missing information and discrepancies before submission. Errors can lead to rework, delays or compliance risks, making validation and human review important parts of the workflow.

PDF in, structured declaration out

Each job starts with a single PDF — typically commercial invoice and related shipping or packing pages. Document understanding runs inside a controlled pipeline with logging and audit copies of raw extraction. Import and export use different schemas and value rules. Monetary arithmetic stays in code, not in the model. When a shipping detail is missing, optional lookup against public carrier tracking may supplement transport fields.

Validation before anyone sends anything

Extracted data is checked against government-oriented master data — countries, codes, formats, required fields. Safe auto-corrections apply where rules allow; uncorrectable errors clear bad values and flag manual review. Jobs with pipeline errors, review flags, or invalid filing-interface output stay blocked. The platform prepares and validates; it does not replace staff judgment or guarantee regulatory acceptance.

Authorised staff stay in the loop

Completed extractions land in a review screen where staff compare the form to the source PDF, edit, and save. Sending to the existing filing toolchain is allowed only when the pipeline succeeded, review flags are cleared, and interface validation passed. Final government submission still happens in the desktop filing workflow Brightex used before this tool existed.

Designed for the client's operating environment

The application is deployed on Brightex's office network and provides an internal workflow for uploading shipment documents, reviewing extracted data and resolving validation issues. Its deployment and data-handling arrangements were selected around the client's operating requirements. The application prepares and validates declaration data, while authorised staff remain responsible for reviewing the results and completing submission through the existing filing software.

All work