Introduction
In the ever-changing realm of Generative Artificial Intelligence, a new tool has popped up. Its scope is to improve learning for Generative AI models, mostly for prompt engineering using Amazon Bedrock's available models. PartyRock it's just a playground that uses Amazon Bedrock, and it aims to let us experiment with different foundation models.
In this blog post, we are going to learn how to use PartyRock with a pretty cool example (my one social media manager friend says it's cool, so I'm sticking with it) 😎.
Social Media Assistant
Essentially, we are going to build a social media assistant that will help us write social media posts starting from a context, which could be a blog post, an announcement, etc. To keep it interesting, we are going to have tone suggestions, word limitations, and a customized post refinement. Yes, we are going to generate images as well 🔝.
Create New Applications
Let's jump right through it and create our own application. Essentially, we need just a great prompt, and the game is done.
Here we can see how we can use the "App Builder" to generate a custom application within PartyRock.
This is my prompt: "This app will help you write better social media posts. By providing the context, the app will generate a social media post. Additionally, we can customize the social media post with our own ideas, and the app will generate a new version of our social media post."
After hitting "Generate App," in a few seconds, we are going to see our app. The app will take a context and generate a social media post. After that, we can customize the generated post with our own ideas and suggestions, and the app will generate an updated social media post.
Adding Widgets
There are five types of widgets we can add:
- User Input
- Static Text
- Text Generation
- Image Generation
- Chatbot
They are pretty much self-explanatory. Let's try to add a few widgets to our application.
With this interface, I'm going to add a widget for selecting a tone and one for limiting the number of words (I'm looking at you Twitter/X). Then I want to add a widget for generating an image related to the prompt so that I can use it in the social media post. You know what could be really handy? Adding a chatbot that gives us feedback on our post ✅.
Results
After spending a few minutes playing with the prompt to find the best one, this is a gif that shows the entire application:
Actually, I see myself using this tool to receive feedback on my own social media posts and improve their quality.
Integration
Now that we've seen what this tool is capable of, I want to mention that we can also customize the single widget in order to:
- Change inference parameters such as temperature and top p
- Change the foundation model used within the widget
This tool is not officially related to AWS, but they built it. When talking with them, they said they are going to monitor the usage and add features to PartyRock as the demand grows. That said, I hope to see some kind of Retrieval Augmented Generation (RAG) widgets as well as image-to-image generation 🚀.
Pricing
As of now, it has a generous free trial. In the future, I expect that we are going to sign up with an AWS Account so that the costs associated will be only on Amazon Bedrock or it will be a subscription to the AWS Marketplace.
Conclusion
Overall, it was a great experience to brainstorm a few ideas and try them out in a matter of seconds. I love the fact that I can reference inputs and outputs from other widgets. I see myself using this tool for testing a few ideas before actually starting to develop them, which means I can see them come to light with little to no effort.
Want to use and clone my application? Here is the link: https://partyrock.aws/u/Depa/cEEGq-AMY/Social-Media-Assistant
If you enjoyed this article, let me know 🚀.
Thank you so much for reading! 🙏 Keep an eye out for more AWS related posts, and feel free to connect with me on LinkedIn 👉 https://www.linkedin.com/in/matteo-depascale/
References
If you want to try it out, here is the website:
Disclaimer: opinions expressed are solely my own and do not express the views or opinions of my employer.