Free 30-minute AI & automation audit with a senior engineer. You keep the notes. Book yours

Use case · Retail & restaurant technology · India

Syncing point-of-sale data into desktop and cloud accounting without per-client code

An in-house integration engine with a database-driven transformation layer, queue-based delivery and a desktop connector for on-premise accounting software.

Client details are kept confidential. This use case describes the problem and what we built.

The brief

This is an in-house integration platform we built and run. Its first deployment was for a point-of-sale and billing software provider whose customers needed sales, purchase and inventory data posted automatically into their accounting systems, whichever ledger they happened to use.

Rather than build a separate integration for every target, we built one engine. Mappings live in the database, delivery runs through queues, and a small desktop connector reaches the accounting software that never leaves the back office.

The challenge

What needed solving

Every business uses a different ledger

Outlets using a point-of-sale and billing platform keep their books in different systems: desktop accounting software on a back-office PC, or cloud ledgers. Each one needs a different payload format.

Desktop software is not on the internet

Popular Indian accounting packages run on local machines and expose only local XML or HTTP ports, so a cloud service cannot call them directly.

Many document types, many mappings

Day-end summaries, sales orders, purchases, vendors, items, warehouses, transfers, credit and debit notes, receipts and journals each need their own field mapping per target.

The solution

What we built

  1. Database-driven transformation engine

    A Spring Boot engine converts incoming JSON into each target's format using mapping rules stored in PostgreSQL (copy, constant, lookup, date arithmetic and nested arrays), with JSON-schema validation and rollback. Adding a mapping needs no code change.

  2. Queue-based, multi-tenant delivery

    Inbound events go through AWS SQS and SNS to platform-specific strategy adapters. A token manager handles OAuth refresh for cloud ledgers, with separate queues, databases and credentials for each environment and tenant.

  3. Desktop connector for on-premise ledgers

    A Windows connector, packaged as a native installer, listens to the cloud queue and writes vouchers and masters into TallyPrime, Busy or QuickBooks Desktop over their local interfaces. It streams results back as it goes.

  4. Provider-neutral messaging library

    A small Java library with a cloud-agnostic publisher interface and separate AWS and Azure modules lets any of our integration apps switch messaging providers without pulling in both SDKs.

  5. Admin console

    A React and TypeScript dashboard manages clients, integrations and connection status.

The outcome

What changed

New integrations by configuration

New document types and targets are added by adding mapping rows and thin adapters, not by writing a new service for each customer.

Cloud and desktop ledgers in one flow

The same event pipeline serves both cloud accounting APIs and software running on a back-office PC.

Reusable across our projects

The engine and messaging library are shared building blocks for our other integration work.

Technology used

  • Java
  • Spring Boot
  • JHipster
  • PostgreSQL
  • Liquibase
  • AWS SQS
  • AWS SNS
  • AWS S3
  • React
  • TypeScript
  • Redux Toolkit
  • Gradle
  • jpackage
  • TallyPrime
  • QuickBooks
  • Xero
  • Zoho Books

Have a similar problem to solve?

Share a few lines about your project. A senior engineer — not a salesperson — replies within one business day.

Free AI audit Let’s talk