🛍️ BuySight

BuySight is a conversational shopping assistant built during a hackathon to simplify and enhance the online shopping experience using AI. The system leverages the Gemini API for natural language understanding and the eBay API for retrieving real-time product listings — providing users with tailored product recommendations directly in a chat interface.

---

🚀 Overview

BuySight helps users plan events or purchase items by simply describing their needs in natural language (e.g., "I'm planning a garden tea party for 15 people"). The system analyzes the message using LLM-based NLP and returns actionable shopping suggestions with direct e-commerce links — all within a clean and interactive chat.

---

🔍 Key Features

• Natural language product discovery via Gemini API
• Real-time product retrieval from eBay API
• Interactive chat interface displaying:
  • Product links
  • Price comparisons
  • Customer reviews
  • Product images
• Future feature (in development):
  • Image composition with DALL·E: Users can upload a photo (e.g., their living room), and the app visualizes selected products within that image — giving a real-life preview before purchase.

---

🧠 Technologies Used

Component	Technology
Frontend	HTML/CSS/JavaScript
Backend	Python + Flask
AI & NLP	Gemini Pro API
eCommerce Data	eBay Shopping API
Image Generation	DALL·E API
Deployment	Localhost / Cloud-ready
Communication	JSON-based exchange

---

🧩 Architecture

The project is divided into five Git branches, each managing a core part of the system:

• GEMINI: Processes user input and generates structured product intent
• EBAY: Queries eBay’s API using Gemini’s output
• CORE: Coordinates backend logic and connects components
• DEV: Frontend interface and integration testing
• MASTER: Stable production-ready version

Each module can be tested independently and communicates via structured JSON objects.

---

📂 Example: Gemini to eBay Flow

Inside the Core/Gemini/Schemes folder, you’ll find example JSON files demonstrating the expected output from the Gemini module. These outputs are transformed into eBay-compatible queries.

Example:

{
  "event": "tea party",
  "recommended_items": ["teacups", "cake stands", "napkins"],
  "budget": "medium"
}

---

🧪 How to Run Locally

1. Clone the repository:

   git clone https://github.com/elalgil/buy-sight-app.git

2. Install dependencies:

   pip install -r requirements.txt

3. Set up eBay API access:
   • Create an eBay Developer Account
   • Generate a User Token
   • Add your credentials to the config file:
     Core/ebay/Entities/ebayconf.yaml

---

🌐 Future Plans

• Add support for more eCommerce platforms (e.g., Amazon, Etsy)
• Enhance multilingual support
• Allow users to pin and save product bundles
• Integrate image generation more deeply into the chat interface

---

🎓 HUJI Hackathon

📅 29.5.25 – Annual Hackathon hosted by the Computer Science School at Hebrew University (HUJI).
🎯 Theme: "The Future of AI in the Next 100 Years"

BuySight was developed during this event as a prototype for conversational AI-driven eCommerce experiences.