ReturnsOS

ReturnsOS is a decision intelligence layer and reverse logistics platform for electronics retail and insurance. When a device enters a processing center, operators rely on tribal knowledge. ReturnsOS standardizes evaluation rules and uses multi-agent AI to maximize the value recovered from every returned asset.

⚡ Input → Output Example

// Input (Device Record)
{ "brand": "Apple", "model": "iPhone 14 Pro", "functional_grade": "C", "cosmetic_grade": "B", "battery_swollen": false, "mdm_lock": "UNLOCKED" }

// Output (Decision Recommendation)
{ "recommended_action": "REPAIR", "estimated_recovery_value": 430.00, "confidence": 0.94 }

🎥 Demos & Links

• ReturnsOS Pitch & Video Script
• System Architecture

🏗️ Architecture Sketch

View Diagram Here

🛠️ Tech Stack

• Frontend: React, Vite
• Backend: FastAPI, Python
• Database: SQLite (PostgreSQL ready)
• Events: Server-Sent Events (SSE)
• AI Provider: Google Gemini Pro

📊 Status

• Implemented: React UI, FastAPI Backend, Rules Engine, Valuation Matrix, Event Streaming, Mobile Flow, Rule Configurator.
• Design-only / Stubbed: Gemini API integration (currently simulated via mock engine for ease of review), RabbitMQ (simulated via SSE).

📁 Repository Structure

• frontend/ - React application
• backend/ - FastAPI application and SQLite DB
• docs/ - Comprehensive technical specifications

📜 License

MIT