We adopted a "Hyper-Agile" automation strategy, leveraging low-code platforms (n8n, Retool) and API-first architecture to bypass the traditional software development lifecycle. We didn't build software from scratch; we composed it from existing powerful blocks.
Week 1: The Data Pipeline (n8n)
We replaced the clipboards with cheap Android tablets running a simple Retool app. Workers scanned QR codes on bins to check items out.
Behind the scenes, an n8n workflow orchestrated the data:
* Trigger: New scan in Retool.
* Action: Instantly update the SQL inventory database.
* Action: If stock < threshold, trigger API call to the ERP requesting a re-order.
* Action: Send Slack alert to the floor manager if critical parts are low.
This eliminated the 48-hour lag instantly. Inventory was now real-time.
Week 2: Visual QA Automation
We installed industrial cameras on the main assembly line connected to a local edge server running a custom Computer Vision model.
* The model was trained on 500 images of "good" parts and 500 images of "defective" ones (scratches, dents).
* As parts moved on the conveyor, the camera snapped a photo.
* The AI analyzed it in <500ms.
* If defective, a pneumatic arm automatically pushed the part into a reject bin, and the system logged the exact time and nature of the defect.
Week 3: The Executive Dashboard
We built a unified Retool dashboard for management.
* Live feed of production throughput.
* Real-time defect rates and Pareto charts of defect types.
* Inventory health status.
This dashboard was the "evidence" required for the client audit.
Deployment & Training
By Day 21, the system was live. Training took 2 hours because the tablet apps were intuitive and resembled consumer apps.
