← Cases
GreeceFinance and insurance2024 – ongoing

National Bank of Greece

AI document classification and information extraction

National Bank of Greece used document AI to reach 90% processing accuracy, Microsoft reported

WorkingVendor case studyVendor ↗

Problem

Branch staff printed and scanned customer documents before sending them to the back office for manual extraction. An earlier rule-based OCR system supported too narrow a range of documents.

Approach

The bank built a document-processing solution with Azure AI Document Intelligence. It classified documents and extracted information, connecting the results to customer-facing systems.

Results

90%document-processing accuracy reported by MicrosoftVendor case study[1]
0.5 secondsprocessing time per page reported by MicrosoftVendor case study[1]

Takeaway

The bank replaced a constrained OCR workflow with AI classification and extraction. Microsoft reported 90% accuracy without specifying the evaluation method.

Categories

Business function
Production and quality
Innovation type
Process innovation
AI technology
Computer vision
Evidence
Vendor case study

Sources

Source summaries

[1]Microsoft5 Jun 2024

Microsoft described the bank's replacement of manual extraction and limited OCR with Azure document AI. It reported 90% accuracy and processing at 0.5 seconds per page.

Full details are in the original sources.

Added 1 Nov 2026 · Updated 1 Nov 2026
Open case page ↗