Autonomous AI Management & Dynamic Repricing Agent for Ozon

A professional modular system for business process automation and financial risk control on the Ozon marketplace (institutional grade). The platform calculates end-to-end unit economics in real time, conducts competitive intelligence (OSINT module), strictly enforces margin limits, and generates search engine optimization (SEO) copy for product cards using built-in AI algorithms.

Read this in other languages: Русский (Russian)

Core Architectural Modules

1. Unit Economics Core (Quant Core): Continuous calculation of net profit, margin, and return on investment (ROI) for each SKU. It factoring in cost price, marketplace commissions, multi-level logistics (FBO rates), and tax rates on the fly.
2. OSINT Monitoring & Dynamic Repricing: Simulates high-frequency scanning of competitor prices. It automatically triggers a price-matching protocol (MATCH_COMPETITOR_LOWER) to keep the product card in the top search results. If competitors drop prices below a critical threshold, the protective HOLD_MIN_MARGIN_LIMIT algorithm triggers, locking the price to safeguard the minimum profit margin.
3. AI Executive Director (Executive Decision Engine): Analyzes inventory turnover and financial performance, generating operational strategic directives for the business owner regarding procurement, warehouse supply chain tracking, and marketing budget allocation.
4. AI Content Factory (SEO Copywriting Factory): Automatically generates marketplace search-algorithm-optimized product descriptions and relevant metadata search tags.

Mathematical Apparatus

The unit economics computational engine utilizes the following objective function to calculate net profit:

$$\text{Net Profit} = P_{current} \cdot (1 - \tau) - C_{cost} - (P_{current} \cdot \mu_{ozon}) - L_{fbo}$$

Where:

• $P_{current}$: Current retail price of the SKU on the marketplace.
• $\tau$: Tax rate (default: $0.06$ for the "Income" simplified tax system).
• $C_{cost}$: Base cost price per unit.
• $\mu_{ozon}$: Dynamic Ozon commission coefficient for the specific product category.
• $L_{fbo}$: Fixed operational expenses for FBO fulfillment, sorting, and trunk logistics.

The lower bound of the safe price ($P_{min}$), guaranteeing the protection of the target profit margin, is calculated as:

$$P_{min} = \frac{C_{cost} + L_{fbo}}{1 - \mu_{ozon} - \tau - \text{Margin}_{min}}$$

Deployment & Initialization

System Requirements

• Installed Docker and Docker Compose.

Quick Start (Build and Run in Background)

Simply copy and paste this command into your terminal within the project directory:

docker compose up --build -d

After a successful build, the interactive dashboard web interface will be accessible in your browser at: http://localhost:8501.

Project Structure

Project Structure

ozon_smart_agent/
├── core/
│   └── analytics.py       # Computational core, OSINT simulator, and AI engine
├── data/
│   └── portfolio_ozon.json # Central SKU registry, cost matrix, and margin limits
├── app.py                 # Control dashboard powered by Streamlit
├── requirements.txt       # Python dependency manifest
├── Dockerfile             # Docker container build configuration
└── docker-compose.yml     # Local infrastructure virtualization orchestrator

Useful Container Management Commands

• Check container status:

docker compose ps

• View real-time application logs:

docker compose logs -f ozon-agent

• Stop and remove containers:

docker compose down

Business Rules Configuration

portfolio_ozon.json

You can modify product parameters, cost structures, competitor IDs, and margin safety limits directly in the configuration JSON file:

{
  "currency": "RUB",
  "tax_rate": 0.06,
  "products": {
    "SKU-IDENTIFIER": {
      "name": "Product Name",
      "stock": 50,
      "cost_price": 500.0,
      "current_price": 1500.0,
      "ozon_commission_pct": 0.15,
      "logistics_fbo": 100.0,
      "min_margin_pct": 0.20,
      "target_competitor_sku": "COMP-ID-01"
    }
  }
}