wAIste: A mobile application that helps with food waste reduction OpenAI technology for sustainability

Tackling the problems of food loss and waste reduction can bring benefits like increasing food availability to the most vulnerable, or boosting productivity and economic growth. And with the right technology, everybody could contribute to a more sustainable environment.

Making a positive impact on the world with the use of new technologies like OpenAI and image recognition is a fantastic cause to get involved in, and I’m always on the lookout for such possibilities.

One way to do this is to take on a project that helps with protecting our planet’s natural resources. That’s why I was thrilled to be invited along with my team to participate in a recent sustainability-themed hackathon.

The hackathon presented us with an opportunity to develop innovative solutions that address important sustainability challenges. We decided to focus on food waste, an issue that affects both the environment and our society.  

Key points:
  • How can technology like OpenAI and image recognition help with household waste management? 
  • How to create an innovative application with an environmentally-friendly goal? 

For the purpose of the event, we played the role of a retail company that wants to create and provide a consumer-facing solution that can help minimize food-related waste.  After 3 intense Design Thinking workshop sessions, culminating in a 3-day hackathon, we have successfully delivered an MVP of the mobile application that we called wAIste.  

Before we go into detail, let me explain why waste management became the focus of our project. 

Why are attempts at food waste reduction so important?

The food waste problem is widely acknowledged as requiring special attention and is notably a part of Sustainable Development Goal No. 12 set by the United Nations. 

Image from The Sustainable Development Goals Report 2022

Image from The Sustainable Development Goals Report 2022 (click to view full-size)

To provide further context about the scale of the food waste problem, consider these statistics.

According to the reported estimates, approximately 931 million tons of food waste were generated in 2019. Of this:

  • 61% originated from households
  • 26% from food service
  • 13% from retail.

Data from the FUSIONS EU report indicate that approximately 88 million tons of food waste are generated annually within the EU. This is equal to:

  • 174 kg per person or
  • 143 billion euros in economic loss or
  • 170,000,000 tones of CO2 emissions.

Per the report, households are the source of the majority of food waste, meaning that waste can be significantly reduced at an individual level. We believe that helping everyday consumers better manage their groceries is a great opportunity to implement meaningful change at scale.

Features of the wAIste application

wAIste uses Artificial Intelligence and Machine Learning capabilities to assist users in reducing food waste and improving garbage sorting, making it easier for them to contribute to a more sustainable environment.

Expiry date monitoring

Our primary objective was to tackle the issue of food waste generated by expired products.

A lot of food is wasted due to spoilage. This can happen for many reasons, such as neglect or forgetfulness. To help solve this challenge, one of the key features of our application was to provide expiry date monitoring and notifications.

The user simply had to provide an image of the groceries they own. Then, the application automatically updates household product inventory and provides estimated expiry dates. This way, people can prioritize which ingredients to use the soonest.

Screenshots from wAIste application showing itemized groceries

The application provides an automatic product inventory and adds estimated expiry dates based on the image

Recipe suggestions

To make the app even more interesting and useful, we added an additional feature using OpenAI capabilities. It suggests complete recipes for the products the app user currently has.

By listing the ingredients, users can access a tab with live recipe suggestions based on what their current inventory.

Screenshot from the application showing generated recipe

The application suggests complete recipes for the products the app user currently has

Taking images of additional groceries expands the range of ingredients and offers more complex and interesting recipe options.

Waste sorting

Another important functionality is product recognition, providing users with guidance on how to sort any product or packaging waste.

The application provides users with product recognition and further guidance on how to sort waste

From the beginning, our priority was to make the application as convenient for the user as possible. Therefore, all input it needs is in the form of images, ensuring effortless use.

No typing is required, and different setups and angles were tested to ensure the best functionality even without perfect lighting.

Subscribe to our newsletter to get more updates like this every two weeks. Sign up

How does the application work?

We developed the application’s functionalities using the following technologies:

  • Mobile application – IONIC, Angular
  • Image recognition – Azure Cognitive Services Custom Vision model
    • Custom ingredient recognition model
    • Custom waste recognition model
  • Backend orchestration – Function Apps
  • Recipe suggestions – OpenAI GPT-3 model
  • Data storage – Azure SQL
  • Image storage – Blob Storage

Being cloud-native in its design, this MVP of the application has great scaling potential. Additionally, the prototype cost is remarkably low, providing a lot of opportunities for further expansion.

This is the overall architecture of the app. It is entirely powered by Microsoft Azure services, which gives it great flexibility both in terms of scale and functionality. Additional services could be easily added or integrated.

Architecture of wAIste application (click to view full-size)

Environmentally-friendly technology comes to the fore

With a proper approach and technological selection, proof-of-concept solutions like this can be delivered rapidly and flexibly. Using the IONIC framework, we have built versions of the app for all major mobile systems and web versions.

The use of Azure Cognitive Services Custom Vision models allows for further model retraining to enhance object detection quality under various conditions. Moreover, the inclusion of Azure OpenAI GPT Models provides unparalleled flexibility in generating responses, requiring only minor adjustments to the prompts.

A simple app with an advanced goal

wAIste can help consumers make the most of their products and reduce waste, improving home economy and looking after the environment at the same time.

The solution could also be used by grocery stores, integrating it into their applications to help their customers manage their purchases better.

Although it’s a standalone customer-facing application by design, its functionalities could also be potentially used by retail companies to enhance their existing apps with new, engaging features.

I am excited about the potential impact of wAIste. At Predica, we are committed to exploring opportunities to further develop and implement this application and we’re proud to be part of the growing movement toward sustainability.

If you’d like to discuss your idea for a solution that helps protect natural resources and mitigate the effects of climate change, get in touch with us today.

Key takeaways:
  1. An application that requires minimal effort from the user can go a long way toward improving household waste management.
  2. An MVP of the application was created within 4 weeks. Using cloud services, it can be easily enriched with new functionalities or tailored to other use cases.

Sign up for Predica Newsletter

A weekly, ad-free newsletter that helps cutomer stay in the know. Take a look.


Want more updates like this? Join thousands of specialists who already follow our newsletter.

Stay up to date with the latest cloud insights from our CTO