Azure-Based Software Product for Intelligent Invoice Data Capture

Customer

The Customer is a US-based provider of supplier relationship management software solutions for mid-market and large businesses that have around 200,000 active users all over the world.

Challenge

The Customer had a cloud product for accounts payable management. To answer their clients’ growing need for further automation of accounts payable processing, the Customer wanted to enrich it with a new software module capable of intelligent invoice data capture. The new module had to work fast and stably even in the context of heavy load conditions and easily scale on demand.

For this reason, the Customer was looking for a trustworthy tech partner experienced in distributed modular software development and AI techs who could take over the full-cycle software product development.

Solution

ScienceSoft’s cross-functional team of a BA, web developers, and QA experts delivered the new module and integrated it with the Customer’s software product hosted on Azure. Invoice processing in the module ran as follows:

• Invoices were sent via integrated software (e.g., a vendor portal) or directly emailed to the system (PDF or scanned documents).
• Invoice data (vendor name, invoice number, date, PO number, description, quantity, unit price, etc.) was extracted with OCR technology.
• Integrated Azure Cognitive Services allowed classifying and structuring the received data and validating its completeness and accuracy.
• Invoices were further routed for approval and payment.

To secure easy scaling and granular improvements of the module, the team decomposed it into decentralized microservices. Azure Service Fabric was used to solve the problems of service communication, discovery, monitoring, provision and update, local testing in the complex microservices-based environment and keep it reliable, highly available and fault tolerant.

Results

The new software module allowed smart paperless invoice processing, improving accuracy and reducing the time spent on data entering and correcting. The Customer increased the loyalty of the existing users and significantly enlarged their business audience. The module demonstrated stable performance in the conditions of a continuously growing customer base.

Technologies and Tools

.Net Core 3.1. React, Azure Service Fabric, Azure Service Bus, Azure Application Insights, Azure Cognitive Services, Azure SQL Database, Azure Virtual Machine Scale Sets, Azure Virtual Network, Azure Application Gateway, Azure Load Balancer, Azure Key Vault, Azure Storage, Azure Content Delivery Network (CDN).