Import & trading · Hong Kong
A trading company
200 supplier invoices a month, keyed by nobody.
Every supplier invoice — scans, photos and PDFs — was re-typed into a spreadsheet by hand. We built an OCR pipeline that turns them into clean, structured rows on their own, with a human touch only where confidence is low.
The challenge
Keying invoices took about two full days a month, and typos in amounts or invoice numbers caused reconciliation headaches that took even longer to untangle.
The solution
Invoices are forwarded to one inbox. The pipeline reads each document, extracts supplier, date, line items and totals, validates the arithmetic, writes clean rows to the ledger sheet, and flags only low-confidence documents for review.
The workflow we built
- One inbox. Suppliers and staff forward invoices to a single address — nothing else changes for them.
- Read & extract. OCR plus AI parsing pulls supplier, dates, line items and totals from scans, photos and PDFs.
- Validate. Totals are recomputed and cross-checked; anything that doesn't add up is flagged, not guessed.
- File & reconcile. Clean rows land in the ledger automatically, ready for monthly reconciliation.
Tools we used
The results
- Two days of keying a month became minutes of reviewing flagged items.
- Amount and invoice-number typos disappeared, so reconciliation stopped throwing surprises.
- Around HK$10,000 of admin time saved monthly — and month-end closes faster.
Representative example — figures illustrative, client anonymised.
Have an operation that looks like this?
Tell us the process and we'll show you exactly how to automate it — free, and no obligation.
Get a Free Automation Audit →