A transaction workflow that keeps the detail

01 — The challenge

Where it started.

The firm processed a large daily volume of transactions and needed a record of each one at a granular level — not a summary, the detail.

Volume and detail pull in opposite directions: the systems that cope with the first usually discard the second.

02 — What we built

The work itself.

A decoupled web application. Built so the parts can be changed and scaled separately rather than as one block.

RabbitMQ carrying the inbound messages. Transactional messages queued from the client’s external application, so accuracy holds when traffic spikes.

SignalR to the user’s inbox. Real-time notification pushed rather than polled.

AWS Cognito and Lambda. Authentication, authorisation and the identity emails handled as managed services instead of hand-rolled.

A daily record, at line level. Reports sent on the day’s transactions, each one with its own detail retained.

03 — Results

What the client has now.

Real-time dashboards giving the firm something to tell its own customers. A per-transaction record that holds up when someone asks about one line. Operating cost reduced by shortening the process rather than by cutting the detail kept.