![Once upon a time... We created an AI Taleteller](https://assets.community.aws/a/2dV737hdmoYhzCrZnEg5LSIIlXq/Tech.webp)
Once upon a time... We created an AI Taleteller
Introducing The Tech Storyteller: an AI assistant to help craft captivating presentations through storytelling techniques
Published Mar 10, 2024
In our quest to promote storytelling for technical presentations, we built "The Tech Storyteller" crafted to elevate the impact of tech presentations with the aid of an AI Assistant.
Our creative process focused on the challenges faced by developers in conveying complex technical concepts to diverse audiences, and we were able to leverage PartyRock's no-code playground environment to easily spin up a first version of the app, and then leverage the different widgets and parameters to customize it.
Community Impact
We created "The Tech Storyteller" with developers and tech enthusiasts in mind. By blending technical details with captivating narratives, it aims to break down communication barriers and make information accessible to a broader audience. Therefore, the potential benefits extend beyond improved presentation techniques, and we see our tool as a way to foster the tech community by facilitating knowledge-sharing.
We created "The Tech Storyteller" with developers and tech enthusiasts in mind. By blending technical details with captivating narratives, it aims to break down communication barriers and make information accessible to a broader audience. Therefore, the potential benefits extend beyond improved presentation techniques, and we see our tool as a way to foster the tech community by facilitating knowledge-sharing.
For our app to better reflect developers' needs, we are keen to listen to feedback and fine-tune the app accordingly. We are optimistic about The Tech Storyteller's potential to become a fun tool for developers' routine use, whenever they are seeking to craft a presentation.
Alternative Development Scenario with Amazon Bedrock
Had PartyRock not been available, we would have built our app with a combination of different AWS services. We could use Amazon Sagemaker as our Machine Learning platform, and create Jupyter Notebook to orchestrate creating a Bedrock client and sending our prompt to a model, and process the response back. The full process is detailed below:
Had PartyRock not been available, we would have built our app with a combination of different AWS services. We could use Amazon Sagemaker as our Machine Learning platform, and create Jupyter Notebook to orchestrate creating a Bedrock client and sending our prompt to a model, and process the response back. The full process is detailed below:
![Alternative Development Scenario with Amazon Bedrock - Presentation and Image Generation A flow diagram of an user interacting with an AI service to generate presentations and an image](https://assets.community.aws/a/2dVMYfe3SGCMH0M3SeWtvsioe7H/Scre.webp)
- [User input] The user provides the main information about the presentation.
- [Generating a presentation with storytelling] With the aid of Amazon SageMaker, we would create a prompt and add the additional parameters (such as model type) to call the Amazon Bedrock service, which manages foundation models, in particular Claude, our Large Language Model of choice to generate the presentation.
- [Response - Text] SageMaker can then handle receiving and processing the text response.
- [Creating the Storyteller's image] We would then make a second call targeted at a Text-To-Image model (Stable Diffusion XL), requesting an illustration of our fantastic Storyteller.
- [Response - Image] SageMaker can then handle receiving and processing the image.
- [Responses to User] SageMaker then returns the responses to the user. Alternatively, to improve the user experience, we could also provide each response (text and image) to the user as soon as it is available, much like PartyRock does.
![Alternative Development Scenario with Amazon Bedrock - AI Assistant Chat A flow diagram of an user interacting with an AI service to ask questions](https://assets.community.aws/a/2dVMUFqA74swFeEgHdnCJNeRULy/Scre.webp)
We had also envisioned a Chatbot widget, so the user can be assisted on their journey:
- [Chatting with the assistant] User would send in their questions, one at a time.
- [Response - Chat] The LLM sends its answers.
- [Response to User] SageMaker then outputs the answer to the User. The cycle continues as the User asks more questions.
One technique we could also leverage is prompt chaining, which is used in conversational AI to create more dynamic chatbots. With prompt chaining, the output of one prompt is added as input for the next, creating a continuous chain that enables context to be always present during the conversation.
It would be ideal to also present the user with a nice front-end. This AWS blog post has a step-by-step on how to leverage an existing UI template and deploy it using Amplify.
With this alternative architecture, the goal would nonetheless remain consistent: to create a tool that empowers developers to craft engaging narratives while conveying complex technical concepts effectively.
In conclusion, our journey with "The Tech Storyteller" reflects a commitment to help developers showcase their work in the best light, and make technical information accessible to all. We see a bright future for our app, and we trust its potential to positively impact the way developers share knowledge and connect with their audiences through storytelling.