Grobo is a tool I created using Google’s AIStudio to vibe code an application that uses AI to help find all the local grocery deals in an area and then creates a menu and shopping list based on the deals. I created it in a couple days using the AIStudio platform. Here was the result, a react-app that uses the Google AI API to find local grocery deals and create a menu and shopping list based on those deals:

I really liked the idea, but I didn’t like the constraints of AI Studio. I wanted to create a tool that could be used by anyone, and I wanted to use the AI models that I wanted to use, not just the ones that were available in AI Studio. So I made a self-hosted version of Grobo that can interact with any the AI of your choice, and that can be run on any platform. This can be found here:

So, using AIStudio I was able to quickly create a fully functional application. The downsides being that it’s a React app, which is cool but not always the best choice, and that it uses the Google AI API. It did allow me to add other AI models in the options with enough coaxing, but it’s not default. The final product was able to find local grocery deals and create a menu and shopping list based on those deals. It was an interesting platform to use, and I learned a few things about how to use AI to create applications.

However, the constraints of AI Studio was too much to make a product that I would want to use, but I really like the idea.

I’m instead going to make it into an open-source tool that I can use to help with my own personal menu planning and grocery shopping, at least. It will be a fun project that will involve a lot of different technologies and it will be a great way to learn about AI and its use in an application.

Design

The application will have a web frontend that will allow users to login and create a profile and setup their preferences. The frontend will be built using Angular. The backend will be built using Go and will use the Chi router. The backend will have a docker image that can be used to run the application locally or on a server, and it will point to a separate AI service that can be used to interact with any AI model, you just point it to the server. I plan on personally testing it with Ollama, but if other people want to support other models it should be easy to do so. The backend will also have a database that will store the user profiles, preferences, and the grocery deals that are found. The database will be built using Pebble, which is a pretty good persistant key-value store that is easy to use and has good performance. I used it with my discord bot and it worked really well.

Front End

The frontend will be built using Angular and will have a simple and clean design. The main page will have a login and registration form (registration should be able to be turned off if the admin wants to manage users manually). It will have a dashboard that will show the user their profile where they can set their preferences for the types of meals they like, dietary restrictions, etc. The dashboard will have a button to scan for new grocery deals or view previously found deals. The dashboard will also have a section for the menu and shopping list that are generated based on the grocery deals and the user’s selections.

When the user clicks the button to scan for new grocery deals, it will send a request to the backend to find new deals in the user’s area. The backend will then interact with the AI service to find the best deals and return them to the frontend. The frontend will then display the deals to the user and allow them to select which deals they want to use for their menu. Once the user has selected the deals they want to use, they can click a button to generate the menu. The menu will be generated based on the selected deals and the user’s preferences. It will contain several meal options for the week for the user to choose from. When the user has selected the meals they want to make, they can click a button to generate the shopping list. The shopping list will be generated based on the selected meals and the grocery deals that are being used. The shopping list will be organized by store and show the user where to find the best deals. Should help save people time and money!

Back End

The backend will be built using Go and will use the Chi router. It will have a REST API that the frontend can interact with to manage user profiles, preferences using a Pebble datastore. The backend will also have an endpoint that the frontend can call to find new grocery deals. When the backend receives a request to find new grocery deals, it will interact with the AI service to find the best deals in the user’s area. The backend will then return the deals to the frontend for the user to select from. The backend will also have an endpoint that the frontend can call to generate the menu and shopping list.