Back to Chronicles
development

QCGenius: Real-time Factory AI Built in One Afternoon 🏭

qcgeniusdevelopmentaimanufacturinggemini-3computer-vision

Today we’re excited to share QCGenius, an app we built in a single afternoon! ⚡

The idea came from the same friend who recommended DashboardGenius. He pointed out a glaring gap in manufacturing: Quality Control (QC) is still shockingly manual.

The Problem: Trust but Verify?

Manufacturers often rely on operators to manually check products and scribble results on paper or enter them into a digital system.

  • The Issue: It’s 100% trust-based. If an operator misses a defect or just ticks a box to get through the shift, the system records it as "Pass."
  • The Consequence: Minor quality slips can snowball into massive rework, product throwaways, and significant financial loss.

The Solution: AI Eyes on the Line

My friend’s concept was simple but powerful: Give operators an iPad.

  • Traceable Record: Instead of just a checkbox, the operator takes a photo.
  • Real-time Feedback: The AI analyzes the image against quality standards instantly.
  • Verifiable Truth: Management gets a visual record, not just a data entry point.

Why Now? The Gemini 3 Advantage 🧠

QCGenius is made possible by the incredible multi-modality of SOTA AI models like Gemini 3.

  • Context Aware: It combines visual data with user input (text & voice recordings!) to understand what it's looking at.
  • Zero-Shot Learning: Unlike old CV models that needed thousands of labeled images, this model knows what to look for without extensive fine-tuning on scarce defect data.

QCGenius Main Menu

Feature Spotlight: Mass SKU Ingestion 📦

One of the coolest features is how we handle data onboarding. Companies have hundreds of SKUs effectively trapped in different formats (ERPs, MES, Excel, CSV).

  • Universal Import: QCGenius can digest product data from massive lists, specific file formats, or even simple screenshots/photos of a product list.
  • Instant Setup: The AI parses the "messy" data and structures it ready for inspection.

QCGenius Product Import

Beyond Products: Work Area Inspection 🧹

My friend also suggested extending this beyond just products to work area inspections.

  • The Goal: Verify clean states, tool placement, and general housekeeping (5S).
  • Voice-First Context: This is where voice recording shines. Even with minimal reference images, the AI can follow verbal instructions ("Check if the wrench is on the shadow board and the floor is clear of debris") just like a human would, validating the scene against the photo record.

What's Next?

We’re thrilled that modern dev tools allow us to build complex, production-ready apps in hours, not weeks. We’ll be rigorously testing QCGenius with our friend after the holidays to get it factory-ready.

Exciting times ahead for manufacturing productivity in the US! 🇺🇸