CoofyAI: An LLM-Powered Intelligent E-commerce Discount Analysis Platform

CoofyAI is a production-grade AI e-commerce discount intelligence platform developed by be-codage and open-sourced on GitHub (released on May 24, 2026). It aims to address consumer pain points in e-commerce promotions: identifying fake discounts, verifying the authenticity of original prices, and distinguishing between marketing tactics and real deals. Core features include intelligent sorting of optimal discounts, fake discount detection, fraud pattern recognition, credibility scoring (0-100), product comparison analysis, and highlighting of the top 5 best deals.

In today's era of rampant e-commerce promotions, consumers often face dilemmas: How to judge whether a "discounted" product is a good deal? Is the "original price" marked by merchants real? Is "limited-time flash sale" a marketing tactic? These problems gave birth to CoofyAI—an AI-driven platform specifically designed to solve such pain points.

Tech Stack: Backend uses Python + FastAPI, frontend is Streamlit, AI model uses Groq API's llama-3.3-70b-versatile, web scraping relies on Selenium + BeautifulSoup, data validation uses Pydantic.

Workflow: User inputs URL → FastAPI receives → Selenium headless Chrome renders the page → BS4 extracts and cleans content → AI Agent analyzes → returns structured data → Streamlit displays.

Core Mechanism: Dynamic web page rendering (handles JS lazy loading), AI-driven analysis (price history analysis, brand reputation evaluation, specification comparison, risk marking). Supported platforms include Amazon India, Flipkart, Myntra, and general e-commerce sites (both single product and multi-product lists are supported).

Application Scenarios: 1. Consumer shopping decision assistance (avoids impulsive consumption and fake promotions); 2. Price monitoring and trend analysis (identifies low-price opportunities); 3. Enhances e-commerce transparency (promotes honest pricing on platforms); 4. Competitor analysis and market research (a tool for merchants/researchers).

Evidence Example: API responses can return structured data including original price, discounted price, discount rate, trust score, and suspiciousness judgment for products (e.g., analysis result for HP Laptop 15s: original price ₹65,000, discounted price ₹45,999, discount rate 29%, trust score 87).

Limitations: 1. Anti-crawler challenges (some websites block headless browsers); 2. Regional restrictions (mainly targeting the Indian market); 3. API dependency (Groq's availability and cost); 4. Insufficient real-time performance (price data may be delayed).

Improvement Directions: Support more platforms/regions, add price history tracking, incorporate user feedback to optimize AI, develop browser extension versions.

CoofyAI is a typical application of AI in the e-commerce field, combining LLM understanding capabilities with web scraping technology to solve real pain points. It not only helps consumers make informed decisions but also promotes transparency in the e-commerce ecosystem. In the future, with the advancement of AI technology, more similar innovative applications are expected to emerge, enabling technology to better serve user needs.