Personalized, AI-Driven Meal Planning

Šta da jedem.ai is an interactive web application I built to help users plan meals, discover new recipes, and optimize grocery shopping—powered by artificial intelligence.
Instead of endlessly wondering “What should I eat today?”, the app delivers concrete, personalized suggestions complete with images and a ready-to-use ingredients list.

Idea & Core Features

Users enter their preferences (meal type, dietary restrictions, target calories, ingredients to include/exclude), and the app generates recipes in real time—with matching AI images and a consolidated shopping list.

Key features include:

• AI recipe generation — GPT model creates unique, easy-to-follow recipes with clear instructions.
• Automatic dish images — generated via DALL·E 3 so every recipe has an appealing visual.
• Weekly meal planner — plan breakfast, lunch, dinner, and snacks for the entire week.
• Shopping list — automatic aggregation of all ingredients from selected meals.
• Email reminders — each morning, users receive the ingredients needed for that day.
• Usage control — a request counter that limits free/guest usage (e.g., 2 free AI queries).

How It Works

1. Preferences input — the user specifies the meal type, constraints, and goals.
2. Content generation — the app requests a recipe and description from the GPT model.
3. Visualization — DALL·E 3 generates a matching dish image from the recipe.
4. Planning — meals are stored in the weekly planner with linked ingredients.
5. Shopping list — all ingredients are aggregated into an interactive, checkable list.
6. Email reminder — a cron job sends daily reminders with the day’s ingredients.

Technologies

• Backend: PHP 8, Laravel Framework
• Frontend: HTML5, CSS3, JavaScript (ES6), AJAX
• Database: MySQL
• AI integrations: OpenAI GPT-4 for recipes, DALL·E 3 for images
• Automation: cron jobs for daily email reminders
• Session & guest logic: tracks users and requests even without registration
• Responsive design: tailored for both mobile and desktop

Notable Implementations

• Slug-based linking — every recipe uses SEO-friendly URLs (e.g., /blog-single?slug=pasulj---prebranac).
• Automatic ingredient extraction — if a recipe lacks a predefined ingredient list, the system parses the recipe content to generate it.
• meal_ingredients table — ensures accurate synchronization between the weekly planner and the shopping list.
• Email template optimization — users receive only the ingredient checklist, without unnecessary clutter.

Results & Impact

By combining modern AI with everyday needs, the application saves time, simplifies planning, and inspires users to try new dishes. It encourages healthier, more varied, and better-organized eating habits.

This project demonstrates how a practical idea, solid system architecture, and AI can come together to create a digital product that is useful, enjoyable, and scalable.

Compact Version (optional)

Overview: AI-powered meal planner that generates recipes, images, weekly plans, and a smart shopping list based on user preferences and dietary goals.
Highlights: GPT-4 recipes, DALL·E 3 images, cron-based email reminders, SEO-friendly slugs, automatic ingredient parsing, and a normalized meal_ingredients schema.
Impact: Faster meal decisions, healthier variety, and a more organized shopping routine.

‍