↓Skip to main content

First Time Dipping my Toes into AI

·534 words·3 mins
Mark Delaportas
Author
Mark Delaportas
ECE student
Table of Contents

This is a work in progress and subject to change as is still in its infancy.

The problem
#

A super market chain needs to embed a product inside their web application that optimizes the grocery shopping experience of it’s customers. In greater detail customers in the retail food industry often struggle with reordering essentials efficiently, leading to missed purchases, stockouts at home, and irrelevant product recommendations. While subscription models exist, they lack true personalization—failing to consider individual shopping frequency, lifestyle preferences (health-conscious, budget-focused, premium), and real-time consumption patterns. In other words there is a need to Develop an AI-driven smart reordering system that predicts when customers need to replenish products based on purchase frequency, household size, consumption patterns, and lifestyle preferences (e.g., budget-conscious, healthy shopper, premium buyer). The system should generate intelligent suggested orders, which customers can approve, modify, or auto-purchase via an omni-channel experience (app, web, or in-store pickup).

Breaking down the problem
#

At first in order for the product to be developed and shipped efficiently the correct development cycle must be chosen. For this I was thinking of choosing extreme programming as it aligns well with sensitive customer data and is also quite agile. Also it shall be an attempt to democratize this sector as most competitive products are only implemented internally for their customers and are very closed source.

Creating Issues
#

To create the correct issues is to do brainstorming in a narrow manner. In greater detail the issues describe the user experience and provide guidelines for the user interface.

As a user I

  • want to see a basket with the products I order most commonly
  • want to see suggestions of probable products I may need
  • want to green light a recurring order with some basic necessities
  • want to chat with an agent to add products to my basket for a specific recipe or a function
  • want to set preferences like budget, calories etc.

However it is also important to define an Administrator role to oversee the user-AI interaction. As an admin I

  • need to check for errors and prevent overflows
  • need to track stagnant orders
  • need to predict demand based on active orders
  • modify supply chains based on price fluctuations while fulfilling customer needs.

Getting more specific a UML Diagram can describe the interface and the interactions. In grater detail to properly make the product the Use Case and Sequence diagrams shall be constructed.

sequenceDiagram actor User participant Web Application participant Backend System participant Database User->>Web Application: Logs in/Opens Grocery List Web Application->>Backend System: Retrieves User's Common Items Backend System->>Database: Queries Purchase History Database-->>Backend System: Returns Frequent Items Backend System-->>Web Application: Displays Common Items in Basket User->>Web Application: Adjusts Quantities/Adds Items User->>Web Application: Requests Product Suggestions Web Application->>Backend System: Requests Product Suggestions (based on history, preferences) Backend System->>Database: Queries Product Catalog & User Data Database-->>Backend System: Returns Suggested Products Backend System-->>Web Application: Displays Product Suggestions User->>Web Application: Adds Suggested Items to Basket User->>Web Application: Initiates Checkout Web Application->>Backend System: Creates Order Backend System->>Database: Updates Inventory & Creates Order Record Backend System-->>Web Application: Displays Order Confirmation Web Application-->>User: Order Confirmation

What is something of a debate; if the Chatbot/Agent shall gain an Actor role in the aforementioned diagrams…