Hello Frasch

Hello Frasch
User Interface

Inspiration

Hello Frasch emerged from the observation that current meal services, such as Hello Fresh, offer allergen filters in a somewhat concealed manner. Recognizing the need for a more intuitive and personalized approach, our project seeks to go beyond by understanding individual preferences. By tailoring recipe suggestions and actively monitoring micro-nutrient levels, we aim to provide a comprehensive solution that not only promotes healthier eating habits but also ensures a more fulfilling and personalized culinary experience for users.

What it does

Hello Frasch operates as a feature to Hello Fresh, offering personalized recipe recommendations for three days a week. Leveraging learned preferences and analyzing previous recipe choices along with associated nutrient levels and deficiencies, the system tailors suggestions to individual tastes and nutritional needs. In the event a user dislikes a specific recipe, they can effortlessly swap it with an alternative recommendation provided by the system to maintain balanced nutritional levels. Continuous tracking of nutrient levels ensures users have real-time visibility into their nutritional intake, fostering a proactive and informed approach to their well-being.

How we built it

We harnessed state-of-the-art LLMs to analyze micronutrient data from the US Department of Agriculture and recipes, creating a robust foundation for our personalized recipe suggestions. Our innovative approach not only aggregates information on nutrient levels but also allows us to craft personalized recipes tailored to individual nutritional needs. The frontend, developed in Vue.js, ensures a user-friendly and immersive experience while providing an insight into the resulting micro nutrient intake.

Challenges we ran into

The probabilistic nature of LLMs created some problems when trying to implement them into a reliable pipeline. Further the API of the US Department of Agriculture was not very stable, which is why a lot of data had to be parsed manually.

Accomplishments that we're proud of

The sourcing of the micronutrient values and matching them to the provided recipe was challenging but it provides an extremely important insight into your personal nutrition. The health benefits that would result from such a feature would be paramount. Further the AI based substitution to satisfy the customers wishes while watching their micro nutrient intake for them is state of the art elevated the product further.

What we learned

The stability of LLMs is an important task for the coming years so that they can be integrated into pipelines effortlessly. Further, when you know what to look for, maintaining healthy micro nutrient levels with a balanced diet is almost easy. That makes it the perfect use case for a product like Hello Fresh to shoulder this task for their customers, helping them live a healthy and happy life.