StackBox AI is a centralized intelligence engine designed for modern logistics and inventory management. The system integrates real-time facility telemetry, predictive routing models, and automated OCR/object detection ingestion into a single operational truth. By combining a high-performance Next.js dashboard with a YOLOv8-powered Android application, StackBox AI enables seamless synchronization between floor operations and managerial decision-making.
Live Deployment: https://se-project-kappa-one.vercel.app/
- Framework: Next.js 15 (App Router)
- Library: React 18
- Language: TypeScript
- Styling: Tailwind CSS 4 with Glassmorphic UI design
- Data Visualization: Recharts for predictive analytics and metrics
- AI Orchestration: LangChain, LangGraph, and Google Generative AI
- Authentication: Google OAuth 2.0 / Supabase Auth
- Platform: Android (Native)
- Language: Kotlin
- UI Framework: Jetpack Compose
- Local Database: Room Persistence Library
- Networking: Ktor Client
- On-Device AI: TensorFlow Lite (YOLOv8) and ML Kit
- Hardware Integration: CameraX API for real-time scanning
- Database: Supabase (PostgreSQL) with Real-time capabilities
- Serverless Functions: Vercel Functions
- Automation: Vercel Cron Jobs for data pipelines
- Communication: Resend for transactional email notifications
- Storage: Supabase Storage for asset documentation
- Real-Time Synchronization: Seamless data flow between mobile scanners and the central dashboard via Supabase Real-time.
- Automated Data Pipelines: Scheduled market data scraping and forecasting via Vercel Cron.
- Role-Based Access Control: Distinct permission levels for Owners, Managers, and Operators.
- Predictive Intelligence: 30-day revenue forecasting and demand horizon visualization.
- Market Positioning: Automated comparison of internal pricing against market trends with AI-driven price adjustment suggestions.
- Neural Data Ingestion: Automated processing of invoices and bills of lading into structured telemetry.
- Inventory Command: Centralized management of clients, employees, shipments, and inventory bundles.
- Stock Risk Alerts: AI-generated warnings for critical and near-exhaustion stock levels.
- Intelligent Scanning: Real-time object detection using YOLOv8 for rapid inventory identification.
- Offline Mode: Robust Room-based local storage ensuring operational continuity in low-connectivity environments.
- QR Logistics: Dynamic QR code generation and scanning for box and item tracking.
- Inventory Management: On-the-go creation and modification of inventory records.
- Gemini 1.5 Flash: Utilized for intelligent invoice parsing, document understanding, and generating predictive market insights.
- Forecasting Engine: Custom regression models implemented via LangChain for demand prediction and revenue estimation.
- YOLOv8 (You Only Look Once): Optimized TFLite model for real-time, on-device object detection, allowing employees to identify assets instantly via the camera feed.
- Google ML Kit: High-speed barcode and QR code recognition for logistics tracking.
- Data Acquisition: Floor employees use the Android application to scan incoming assets. YOLOv8 identifies items, and the app records telemetry.
- Cloud Integration: Records are instantly synced to the Supabase cloud layer, triggering real-time updates on the Dashboard.
- Automated Processing: Vercel Cron jobs trigger at scheduled intervals to ingest market data and run the predictive intelligence pipeline.
- Managerial Analysis: Dashboard users review AI-generated forecasts, stock risks, and market gaps.
- Operational Action: Owners utilize AI recommendations to adjust pricing, reorder stock, or reassign logistics resources.
| Dashboard Overview | Predictive Analytics |
|---|---|
| [Placeholder: Dashboard Screenshot] | [Placeholder: Predictions Screenshot] |
| Scanner Interface | Inventory Management |
|---|---|
| [Placeholder: Scanner Screenshot] | [Placeholder: Inventory Screenshot] |
- Navigate to
StackBox-Dashboard. - Install dependencies:
npm install. - Configure environment variables in
.env.local. - Run development server:
npm run dev.
- Open
StackBox-EmployeeAppin Android Studio. - Sync Gradle dependencies.
- Configure
local.propertieswith Supabase credentials. - Build and deploy to an Android device.
The web dashboard is optimized for deployment on Vercel, utilizing Vercel Edge Functions and Cron Jobs for maximum performance and reliability.
Copyright 2026 StackBox AI Architecture. All rights reserved.