InventoPlate

Inspiration

As college students that also love experimenting with cooking, we know how painful it can be to spend time finding a recipe, then making a list of ingredients, then manually purchasing each of those items. It is a process that takes away from the relaxation and peace that cooking can bring. Fully-stocked pantries are a myth for busy people. So what can someone that likes cooking but also has a life do?

What it does

InventoPlate automates the process of looking up a recipe, finding the ingredients, adding them to cart, and then buying them into an easy solution. Simply tell our web app what you're in the mood for, and it'll provide you with recipes with in-stock ingredients.

When you're on a Square Online site that uses InventoPlate, a Snippet will show up asking the user if they want to optimize their shopping process.

How we built it

We used the Next.js framework to build our web app. With server-side rendering it gives better performance. It is also easy and fast to develop with and deploy on hosting platforms like Vercel.

For the front-end and components we used Material UI as it is a fast and easy way to build a robust UI. This freed up valuable time for the back-end integration with GPT.

For the Generative AI model we used, we tried and tested several OpenAI models like babbage, davinci etc. and found gpt-3.5-turbo to be the best one in terms of accuracy. We trained the model to provide us with recipe data.

In order to integrate this with Square services and provide value, we used Inventory API to check for in-stock ingredients and to adjust the inventory after purchase. We used Snippets API that would link to our service and a dummy Square Online site to act as a grocery store. We used Checkout API to process the transactions.

Challenges we ran into

We had some issues with the scope of Snippets API. We wanted to implement our whole React app into a snippet using Snippets API. However, it turned out to be harder than we had imagined. It felt as though we were trying to push beyond the scope of Snippets API.

Accomplishments that we're proud of

What we learned

What's next for InventoPlate

Ideally our service would be an easy add on to any Square Online merchant site where the user can use a chatbot like service to make purchase decisions. Further use cases could include: suggestions for PC building hardware, recipes and grocery, furniture. We essentially want to add value to each Square Online merchant website so the user is never confused about purchasing items.

Built With

• openai
• react