Case study · Media and Entertainment
Iyuno cut invoice processing from days to minutes.
Native helped Iyuno build an extraction engine that reads the invoices reaching its finance team, whatever the format, at 99.9% field-level accuracy. It gave the team back over 250 hours.
Iyuno reached 99.9% field-level extraction accuracy, and human extraction errors dropped to zero. Invoice processing went from days to minutes.
About Iyuno
Iyuno is a global leader in localization, dubbing and media services, supporting the world's largest entertainment companies across film, television and streaming. With 30+ offices worldwide, it handles high-volume intercompany transactions across regions, currencies and formats.
The challenge
About 400 invoices arrived each month as PDFs from entities around the world, each with its own layout. Finance staff read every one by hand to pull out invoice numbers, PO numbers, line-item amounts and currencies, then checked them against internal systems.
Processing took hours per batch and days end to end, and human extraction errors reached downstream systems. The work couldn't grow without adding headcount.
A template-based OCR script that only handled clean invoices wouldn't hold. The variety of layouts and line-item structures needed a system that could read any format.
What we built
How it works
- 01IngestAn invoice lands in the monitored inbox and is split into page images.
- 02ReadBoth OCR engines extract the fields, keeping each value's position on the page.
- 03ReconcileGemini Pro compares the two readings and validates the result.
- 04Match or escalateClean invoices are matched against internal systems and move on. Anything anomalous is routed to a person with the details.
Results
The engine reclaimed over 250 finance hours, freeing the team for higher-value financial work. It runs as a production component of Iyuno's finance infrastructure, with every result auditable.
