AI Recipe Creator

Inspiration

Multiple people have different dietary restrictions or preferences that limits their recipe selection. With that in mind, I wanted to created an AI chatbot that curates recommended recipes based on the user's preferences and restrictions.

What it does

Our team created a AI chatbot using Amazon Bedrock large language models to log in user preferences and dietary restrictions for their personal recipes. If the user likes the recipe, then the agent will download the recipe to their local files and log in memory for future references. In addition, the agent will also be able to recommend similar recipes.

How we built it

With python, the user interface is created using streamlit package. Memory logging was done with AgentCore strands integration to prioritize user preferences outlined in the system prompt.

Challenges we ran into

There were multiple ways to initialize and log memory, dropping and retrieving files from our s3 bucket service, connecting streamlit to AgentCore, generating, retrieve, and reading files from and to local cache.

Accomplishments that we're proud of

-Memory caching based on the user's preferences and dietary restrictions. -Agent able to accept the configured payload coming from streamlit. -Generating, retrieving, and reading files from local cache if the user likes the recipe.

What we learned

-Creating custom tools and apply it to our Agent. -Memory logging for short and long term. -Streamlit payload to configure memory and Agent. -File creation, saving, retrieving, and reading of recipes generating by the agent.