AI-Powered Daily Supermarket Price Extraction Pipeline for the Saudi & GCC Market

This project is a demonstration of the architecture and flow logic I designed for the backend of an AI-powered WhatsApp shopping assistant, targeting the Saudi and Persian Gulf markets.

The concept is straightforward: every day at 3:00 AM, an automated workflow triggers in n8n. It uses Apify to scrape daily prices and promotional flyers from major retailers like Othaim, Panda, Sary, Amazon, and Noon. It then passes this raw, messy data through GPT-4o-mini to clean and standardize it (since every store formats its data differently), and finally saves it into Supabase, ready for the WhatsApp assistant to query in real-time.

What you see in the image is Workflow 1 of 3 of the complete system. The other two workflows are designed to be:

A real-time handler that receives shopping lists via WhatsApp (text, audio, or receipt photos) and replies in under 5 seconds with the best prices in an authentic Saudi dialect.

An integration with local payment gateways like Tap or Moyasar to handle subscription activations.

I built this flow to demonstrate exactly how I would structure the project's complete backend: which nodes to use, the data sequence, how to handle transformations between steps, and how to optimize the pipeline to meet strict timing requirements.