Stock counts ten times faster with QR-labelled parts.

Denver Auto Spares strips five different car brands in Port Elizabeth and resells the parts to walk-in, phone and online customers. Off-the-shelf retail and accounting tools don't model what they actually do — a single car becomes a long list of individually-tracked parts, each with photos, pricing and a status (available, in cart, sold, damaged, lost).

Stock counts dragged at the end of every shift. Counter staff captured customer details on paper, then re-typed them into the system, costing roughly twenty minutes per new customer and introducing typos that resurfaced months later on invoices.

Denver Auto Spares brought us a written spec broken into numbered work orders — WO1 through WO11 — and asked us for a bespoke build rather than an automation layer. We worked through the work orders sequentially against a single .NET 8 codebase rather than starting greenfield each time, so every release built on the last.

The app is a custom ASP.NET Core MVC application running on one AWS Lambda function fronted by API Gateway, with MariaDB for storage. The front-end is server-rendered Razor with vanilla JavaScript per page — no SPA framework. Where third-party libraries earned their place, we used them: QuestPDF for invoices, quotes and credit notes; QRCoder for printable part labels with a batch print queue; SixLabors.ImageSharp to resize and thumbnail every part photo on upload; Twilio for the staff PIN login.

We built the public customer-registration form to write directly into MariaDB with input validation and bot protection, removing the paper-and-retype step at the counter. The bank-statement CSV import ships with duplicate detection so a re-uploaded month does not double-book.