PayNex ⚡

Autonomous Agentic Commerce Ecosystem with "Aura" AI Shopping Assistant, Biometric Authorization, and Privacy-First Merchant Auditability.

---

🌟 Executive Overview

As autonomous AI shopping assistants evolve, they face two fundamental challenges:

1. Unbounded Financial Risk for Buyers: AI agents initiating payments without explicit human-in-the-loop consent.
2. The Merchant "Black Box": Merchants having zero visibility or telemetry into why an AI recommended one product over another.

PayNex solves both sides of the transaction:

• On Mobile (Buyer): Buyers converse with Aura, an elite AI shopping assistant built on Google Gemini 3.1 Flash-Lite. Aura handles multimodal visual discovery (uploading product photos), natural language product recommendations, and memory across chat sessions. Every checkout is strictly protected by a Native Android Biometric Security Gate (Fingerprint / 4-Digit PIN) before initiating Razorpay payments.
• On Desktop (Merchant Studio): Merchants receive a Live Telemetry & Audit Dashboard logging vector similarity scores, RAG retrieval decisions, security gate passes, and captured Razorpay transactions in real time—with privacy-first buyer query masking.

---

🏗️ Architecture & Ecosystem

graph TD
    subgraph Mobile Buyer App [Android App - KMP]
        UI[ChatGPT-Style UI & Aura Chat]
        VS[Multimodal Visual Search]
        BG[Biometric / PIN Security Gate]
        RZP[Razorpay Android SDK]
    end

    subgraph Desktop Merchant Studio [Desktop App - KMP]
        MS[Inventory Management & CSV Upload]
        PG[Pagination & Catalog Search]
        LT[Live Telemetry Audit Trail]
        TD[Transaction Drill-Down Details]
    end

    subgraph Backend Brain [FastAPI Service]
        API[REST API Endpoints]
        GEM[Gemini 3.1 Flash-Lite & Embedding-2]
        RAG[Multimodal Vector RAG Engine]
        DB[(Persistent Catalog & Audit Store)]
    end

    UI --> API
    VS --> API
    BG --> RZP
    MS --> API
    LT --> API
    API --> GEM
    API --> RAG
    RAG --> DB

Loading

---

✨ Key Features

🛍️ Mobile Buyer Experience ("Aura")

• ChatGPT / Gemini Style Conversational UI: Multi-session management, sidebar conversation history, "New Chat" workflows, quick suggestion cards, and auto-generated session titles.
• Multimodal Visual Search: Upload product photos (e.g. sneakers or watches) to find matching catalog items using Gemini vision inspection combined with vector RAG.
• Human-in-the-Loop Security Gate: Every purchase requires explicit local authorization via Android BiometricPrompt or fallback 4-Digit PIN.
• Razorpay SDK Integration: Seamless test-mode payment creation upon biometric authorization.
• Full-Screen Image Zoom: Tap any attached image thumbnail to view a high-resolution modal preview with close controls.
• Session & Message Disk Persistence: Chat sessions and message history persist locally across process restarts.

📊 Desktop Merchant Studio

• Minimalist Stripe/Linear Style UI: Charcoal #0E0E10 and Gold #D4A85A aesthetic with subtle status indicators and custom window decorations.
• 250+ Product Inventory Management: Supports large catalog uploads via CSV with smart merging and stock updates.
• Catalog Pagination & Search: Real-time filtering by Name, SKU, or Category, with page navigation (10, 25, 50 rows per page).
• Live Telemetry Audit Trail: Inspect real-time agent decisions, similarity match percentages, and RAG retrieval logs.
• Transaction Drill-Down: Click any transaction row to open a full breakdown dialog detailing payment status, order references, SKU matches, and security gate status.
• Privacy-First Masking: Buyer raw queries are masked in merchant audit logs while exposing explainable recommendation metadata.

---

📂 Project Structure

PayNex/
├── androidApp/          # Android Mobile Application (FragmentActivity, Biometrics, Razorpay)
├── desktopApp/          # Desktop Merchant Studio App (Compose for Desktop, Custom Dark Window)
├── shared/              # KMP Shared Module (Compose Multiplatform UI, ViewModels, Ktor Client)
│   ├── src/commonMain/  # Common UI (BuyerScreens, MerchantScreens, AuraChatInput, AuraSidebar)
│   ├── src/androidMain/ # Android Platform Insets, Disk Storage, Image Pickers
│   └── src/jvmMain/     # Desktop Platform File Storage, AWT Integration
├── backend/             # FastAPI Brain (Gemini 3.1 Flash-Lite, RAG, Persistence, Razorpay)
│   ├── main.py          # FastAPI Endpoints, RAG Engine, CatalogStore
│   └── catalog_store.pkl# Persistent In-Memory Database (Pickle)
├── build.gradle.kts     # Root Gradle Build Script
├── settings.gradle.kts  # Project Settings
└── README.md            # Master Project Documentation

---

🛠️ Tech Stack & Dependencies

• Language: Kotlin 2.x & Python 3.10+
• Frontend: Kotlin Multiplatform (KMP), Compose Multiplatform 1.11+
• Mobile SDKs: AndroidX Biometric, Razorpay Android SDK
• Backend Framework: FastAPI (Python), Uvicorn
• AI Models: Google GenAI SDK (gemini-3.1-flash-lite for chat/vision, gemini-embedding-2 for vectors)
• Networking: Ktor Client (OkHttp for Android, CIO for Desktop)
• Serialization: kotlinx.serialization
• Image Loading: Coil 3 (Compose Multiplatform)

---

🚀 Getting Started

1. Backend Setup

# Navigate to backend directory
cd backend

# Create and activate virtual environment
python -m venv venv
# On Windows:
.\venv\Scripts\activate
# On macOS/Linux:
source venv/bin/activate

# Install dependencies
pip install fastapi uvicorn google-genai pandas numpy razorpay python-dotenv pydantic

# Create .env file with your API keys
echo GEMINI_API_KEY="your_gemini_api_key" > .env
echo RAZORPAY_KEY_ID="your_razorpay_key_id" >> .env
echo RAZORPAY_KEY_SECRET="your_razorpay_key_secret" >> .env

# Run FastAPI server
python main.py

Server runs on http://0.0.0.0:8000

2. Running Android App

# Build and run Android debug app
./gradlew :androidApp:assembleDebug

Note: If running on a physical Android device, update baseUrl in shared/src/androidMain/kotlin/com/paynex/app/Platform.android.kt to your PC's local IP ( e.g. http://192.168.1.XX:8000). For emulators, use http://10.0.2.2:8000.

3. Running Desktop Merchant Studio

# Run Desktop application
./gradlew :desktopApp:run

---

📄 Documentation Modules

For detailed module-specific guides, see:

• 📱 Android App Documentation
• 💻 Desktop App Documentation
• ⚙️ Backend Engine Documentation
• 🧩 Shared KMP Module Documentation

---

📜 License & Acknowledgments

Built for Track 01: AI Growth & Agentic Commerce (Razorpay Buildathon).
Crafted with ❤️ by Vedant Kakade.