Between 1 and 7 June, FoxID passed a milestone we have been working towards for years: more than 50,000 identity documents processed in a week.

We achieved a 99.2% document processing success rate and of those subsequently 98.1% field extraction rate across seven days of continuous, real-world processing. For an Australian-built OCR platform reading driver licences in every condition the field can throw at it, that is a week worth marking.

A milestone is only as meaningful as the quality behind it. Volume without accuracy is just throughput. What made this week stand out was that the platform got faster and more reliable as the load went up.

• Over 50,000 transactions processed (1–7 June) – a 99.2% document processing success rate
• 5.8 seconds median processing time per document

These are not lab numbers. They are aggregated across more than 50,000 live documents from real customers, captured on real scanners, in real lighting.

The headline volume matters, but the field-level extraction quality is what our customers actually rely on. Across the week, the fields that drive verification decisions performed at the top of their range:

• Country: 99.85%
• Document Type: 99.83%
• Last Name: 98.41%
• First Name: 98.35%
• Face Detection: 98.20%
• Date of Birth: 97.26%
• State: 97.15%
• Signature Detection: 95.68%

For an average KPI field extraction rate of 98.1%

The reason FoxID held its response times steady under record load comes down to architecture. FoxID Gen 5 introduced a parallel processing pipeline, and this week it really earned its keep.

After the sequential stages (OCR and document classification) complete, FoxID runs portrait detection, rotation correction and signature detection in parallel rather than one after another. Over the week, that parallelisation delivered a measured 24% reduction in total processing time per document. It is the difference between a platform that slows down when it gets busy and one that simply absorbs the load.

The whole pipeline — from authentication and EXIF handling, through OCR and classification, to storage and validation — is built and hosted by IDV Pacific in Australia.

FoxID began in 2019 because the global OCR providers could not keep up with Australian identity documents — slow template updates, licensing that did not fit local infrastructure, and support time zones on the other side of the world. We built our own, template-free engine instead, one that reads a licence by understanding its layout rather than matching it against a fixed template.

Five generations later, a week like this one is the payoff. More than 50,000 documents, processed in seconds each, at 99%+ accuracy on the fields that matter, entirely on Australian soil. The platform that started below 90% accuracy in its first generation now clears the high-90s at a scale we could not have run two years ago.

To every customer who sent a document through FoxID last week: thank you for the trust, and for helping us cross a line we are proud of.

If you would like to see how FoxID performs against your own document mix, request a demo and we will walk you through it.