Hero image coming soon
001 — CASE STUDY

Vending operations, in production.

An AI-powered mobile app that turns any smartphone into a vending operations scanner.

CLIENTPackStar
YEAR2025
ROLEProduct design, engineering, AI systems architecture
STATUSLIVE
AI SYSTEMSCOMMERCEHARDWARE

Vending operators manage hundreds of machines across dozens of locations. Every restock visit requires a manual inventory count, handwritten notes, and spreadsheet reconciliation. The process is slow, error-prone, and disconnected from purchasing decisions. PackStar needed to collapse that entire workflow into a single phone scan.

We built a React Native app with an AI vision pipeline — the operator points their phone camera at a vending machine, and the system identifies every product, counts quantities, flags low stock, and generates a restock order. The AI runs a multi-stage pipeline: image capture, product recognition via a fine-tuned model, inventory reconciliation against the machine's planogram, and automatic purchase order generation. The entire scan-to-order flow takes under 90 seconds.

From camera scan to purchase order in under 90 seconds, with near-perfect product recognition across 200+ SKUs.

<0sSCAN → ORDER
0%SKU ACCURACY
0moIN PRODUCTION
TECH STACK
  • React Native
  • Expo
  • Supabase
  • Claude Vision API
  • RevenueCat
  • EAS
CREDITS
Founder & LeadBrian Sica

Point. Scan. Restock. No clipboard required.

Studio note
NEXT — MORE WORKAll work →