Introduction 

The client is a leading expert in B2B payments focused on enabling suppliers to optimize their card acceptance by lowering costs, simplifying and expediting reconciliation.

Business Objective

• To create a scalable, reliable and automated platform that simplifies the use of commercial cards for the entire payment's ecosystem.

• To extract information from vast unstructured data residing in emails and excel sheets.

• To create middleware to process settlement transactions.

• Structured, Semi-structured data with transformations in multiple locations.

• Constant data streaming with delta changes

• Not a well architected existing system

• Development of a cognitive data lake to manage data from multiple sources and facilitate transaction processing

• Automation of inputs for existing Suppliers/ Gateways

• Migration of historical data with Real-time data repository in NoSQL

• APIs to integrate 3rd party software with data engineering platform

• Data validation between sources and Data Lake

• Automation and scheduling of reports & GUIs for business usersTechnological Framework

Impact

• A multi-access architecture with a single source of truth was designed.

• An automated, secure and seamless transaction processing solution was developed with real-time alerts.

• Transaction details were crawled from ~4,000 mailboxes with different mail templates.

• Processes were created to generate payout files to concerned stakeholders.

Why this setup?

C# and Python together power the automation of transaction crawling, transformation, and payout file generation, while NiFi or Kinesis handles the business logic for processing records from ~4,000 mailboxes.

Why this setup?

A dual storage approach — SQL for structured settlement records and NoSQL for real-time transaction ingestion — provides both the consistency needed for financial data and the flexibility required for unstructured mailbox inputs.

Why this setup?

The multi-source integration framework ensures all supplier inputs, gateway feeds, and mailbox data are automatically ingested, validated, and processed with real-time alerts for exceptions and approval workflows.

Takeaway

Zapcom built a cognitive, automated B2B payments processing platform that eliminated manual data extraction from ~4,000 mailboxes, established a single source of truth, and enabled fully automated, real-time settlement and payout generation.