My First AI App with Amazon PartyRock

I've been wanting to try Amazon PartyRock for a while, mostly to see what I could actually build with it. I didn't have a huge project idea in mind. I just thought of something simple that I would actually use.

You know that moment when you open the fridge and find a bunch of random things sitting inside, but have absolutely no idea what you can make with them? That was basically my idea. What if I could just enter whatever ingredients I had and get a recipe out of it?

The Idea

The basic idea was simple: enter the ingredients I have, choose the kind of food I want, and mention how much time I have.

I started by describing this idea in a prompt, and PartyRock gave me a starting point that I could experiment with.

My app ended up with three main inputs:

Pantry Ingredients

Preferred Cuisine Style

Time Available

These are used to generate a recipe based on what I enter.

I also added a Plated Dish Visual to generate an image of the recipe and a Live Chef Assistant so I could ask follow-up questions about it.

How Easy Was It?

Honestly, this was my favourite part of using PartyRock.

I didn't have to start by setting up a project or writing a lot of code. I could describe what I wanted and immediately start experimenting with the result.

Once I had the basic version, I could change the inputs, adjust the prompts, and add things as I got new ideas.

I started with:

Ingredients → Recipe

and eventually turned it into:

Ingredients + Cuisine + Time → Recipe → Dish Visual + Chef Assistant

Seeing that happen just by experimenting made the whole process feel surprisingly simple.

What I Learned

The app also gave me a practical way to understand how much context matters when working with generative AI.

For example, asking:

> “What can I make with eggs and onions?”

is very different from telling it the ingredients, the cuisine I prefer, and the amount of time I have.

I could actually change those inputs and see how the output changed, instead of just reading about prompting.

That was probably the most useful part of this experiment for me. I wasn't trying to build something complicated; I was just trying different things and seeing what worked.

This was a small experiment, but it made me curious about what I could try next.

I want to explore Amazon Bedrock to understand more about what's happening behind generative AI applications, and I'm also interested in trying Kiro and seeing how it fits into my development workflow.

For now, PartyRock was a pretty fun way to take a random idea and turn it into something I could actually interact with.