Inspiration

With my wife, the chef at home, away on a four-week trip, I found myself confronted with a dilemma. Ingredients in the refrigerator were in danger of expiration, and I was determined to avoid the wasteful act of discarding them. In a stroke of inspiration, I had an idea — to harness the power of an AI assistant to transform the ingredients in my fridge into tasty meals.

What it does

By inputting my food inventory, the app will suggest three (could be more) recipes with detailed cooking steps and calculate the calorie count of each meal. It will also generate pictures of the meals, allowing me to select my favorite one and follow the steps to prepare it.

How we built it

I built it on PartyRock that is powered by the foundation AI models from Amazon Bedrock. By using the widgets provided in PartyRock APP design GUI page, I could easily create the user input controls, and use prompts to guide AI models to generate the text and image outputs

Challenges we ran into

The generated images can vary in quality, and to improve the consistency, more clear prompts and fine-tuning of the parameters are necessary.

Accomplishments that we're proud of

With this app, IT guys without the cooking skills like myself can effortlessly cook by following the recipes and savor the delicious meals we prepare.

What we learned

We learn how to write prompts and tweak the classifier-free guidance (CFG) scale (AI Model parameters) for better image generation quality by. How to compile your know-how to prompts will be the key to build a generative AI APP powered by LLM.

What's next for Cook What's In Your Fridge

PartyRock did not support file import or external link currently to extend the preferred content. It would be better that users could import cookbooks with the preferred recipes and guide AI models to use the imported data.

Built With

  • partyrock
Share this project:

Updates