Plan Once, Shop Once

An AI meal planner for busy families. Plan weekly meals, generate grocery lists, and turn natural language into structured recipes—saving time on planning and more time with family.

Comment

Inspiration

Four months ago, my wife and I welcomed our first baby. Suddenly, time became our most limited resource.

Before, I could casually browse Costco and decide what to cook for the week. Now, I wanted to plan meals as efficiently as possible, finish grocery shopping in one trip, and spend more time with my family.

That's why I built Plan Once, Shop Once.

What it does

Plan Once, Shop Once is an AI-powered meal planning app (currently available as a Progressive Web App).

Users can:

  • Create recipes using natural language with AI.
  • Plan weekly meals from their recipe library, or let AI generate meals based on their preferences.
  • Automatically consolidate ingredients into a single shopping list.
  • Review recipes and cooking steps while cooking.

Instead of juggling recipes, grocery notes, and meal ideas across multiple apps, everything lives in one place.

How we built it

I designed the overall product vision, user experience, and system architecture.

The app is built as a React-based Progressive Web App, deployed on Vercel, with Supabase handling authentication and database services. The long-term plan is to package it into a native mobile app using Capacitor.

Rather than writing every line of code myself, I used Codex and other AI tools as my full-stack development partner. I started with wireframes and product requirements, then worked with AI to build the frontend, backend, database schema, and application logic. Development was highly iterative - I continuously refined features through conversations with AI until the product matched my vision.

Challenges we ran into

Building with AI introduced a different set of engineering challenges.

  • Planning across long tasks. Codex can generate a solid implementation plan, but it doesn't always follow that plan consistently throughout a long development session.
  • Maintaining context. During larger refactors or debugging sessions, fixing a later issue—such as a database error - sometimes caused Codex to unintentionally revert earlier UI improvements.
  • Architecture still requires humans. AI is excellent at implementation, but it still depends on developers to define the overall system architecture, technology stack, and product direction.

These experiences taught me that AI is an exceptional coding partner—but successful projects still require strong product thinking and technical guidance.

Accomplishments that we're proud of

As someone without a software engineering background, I'm proud that I turned an idea into a real product using AI. More importantly, this isn't just a demo - we actually use it every week to plan our family meals and grocery shopping.

What we learned

This project completely changed how I think about software development.

AI can reliably implement many technical details, but building successful products still depends on humans making the right architectural and product decisions.

I also learned that communicating clearly with AI is a critical skill. The better I could describe a bug or a desired behavior, the better AI could help solve it.

Going forward, I believe developers will spend less time writing code and more time designing systems, defining product direction, and collaborating with AI.

What's next for Plan Once, Shop Once

Next, we plan to:

  • Package the PWA into a native iOS and Android app using Capacitor.
  • Integrate with online grocery platforms like Instacart, enabling one-click shopping directly from a meal plan.
  • Build a recipe-sharing community where users can discover, save, and remix meals from others.

Our long-term vision is simple:

From "What should we eat?" to groceries at your door—with AI handling everything in between.

Built With

Share this project:

Updates