After having kids, my attitude towards everyday tasks changed. I needed quick, efficient, and easy for everything because managing littles is extremely time consuming. My problem was with planning meals and snacks for the week. Every meal planning app on the market either didn't let me see things at a weekly level, or I couldn't incorporate snacks in addition to meals.
Sure, I could just write this all down in a notebook, but that wasn't a sustainable option longterm. I even tried coming up with a solution in Figma where I made components and variants out of my meals and created a weekly menu. That wasn't enough, though.
So...I decided to take my shot at vibe coding and create my own app. I didn't think it would be that complicated to do, but I quickly realized how much underlying logic an app like this actually needs:
-
How would I set up user authentication so that meals are saved?
-
What if the user wants to track their macros, too?
-
and so on.
My choice in software came after trial and error. I initially started with Replit and then Leap and ultimately wasn't a fan of either primarily due to the prompting style required and the amount of control given to the user, or lack thereof. I thought something like Cursor or Windsurf would be too technical for me. Turns out, I need to give myself more credit.
I finally went for Cursor, then had to make the decision of which LLM to use. I was torn between GPT-5 Codex and Sonnet 4.5 so I decided to use Cursor 2.0's feature of running both at the same time to see which was better. I noticed myself starting to think more along the lines of a software engineer when I spotted the difference in lines being added and total lines of code. I wanted to make sure I was keeping the code as clean as possible with minimal noise.
I started with Plan Mode asking the agent to help me create a weekly meal planning app that would allow users to save particular recipes for reuse and see their weekly menu from a snapshot POV. The good thing about plan mode is that it asks questions you sometimes forget to think through.
The first of these questions was what tech stack I wanted to use and the second was how I wanted to set up user authentication and what database I wanted to use.
At this point, I went ahead and set up a git repository for version control.
Sonnet 4.5 was my main coding partner: it was very attentive to what I wanted, and seemed to fix bugs relatively quickly.
I used Vercel to host, instantDB for my database, and ExpoGo to test.
After finishing the first working version, I started using it myself for everyday planning. I quickly grew tired of having to run commands in my terminal and having to scan a QR code to get into ExpoGo to use the working app, so I decided to submit to the App Store.
A few submissions later to work out some kinks, Plated was finally on the App Store. I've since been using the app every week to help with planning and have even gotten a few friends to be test dummies to help me make it better.
