Case Study: NDA Company validates cost-efficient OCR with The Software House

NDA Company validates custom OCR in 2 months with The Software House

The Software House was engaged by a major reinsurance company (referred to as NDA Company) facing a challenge with processing a high volume of poor-quality scanned documents, including handwritten receipts. Their existing external OCR tool was becoming too expensive due to its usage-based pricing model, and they needed to decide whether to commit to a long-term contract with that provider or develop a custom, fixed-cost solution to extract and classify critical data.

The Software House conducted a two-month R&D project to validate a custom OCR approach using Python, OpenCV, and TesseractOCR. While the proof-of-concept achieved 85-90% accuracy on machine-printed text and reduced processing times by 31% through image optimization, the research concluded a custom build would not be cheaper for the client's current scale. Based on this data-driven finding, The Software House recommended and helped the client implement Azure AI Document Intelligence, which offered a lower overall cost due to preferential startup pricing and seamless integration with their existing Microsoft cloud infrastructure.