Aesop AI

Create interactive, illustrated storybooks by choosing the path of the story in real time.

Comment

Inspiration

We came up with the idea through our interests in AI's ability to reveal to us unique and novel perspectives. Just as storytelling unveils unique thought and perspectives, we wanted to test how AI might unfold narratives in innovative ways. Therefore, our goal was to create a product that reveals many perspectives, while at the same time provide the user with an immersive experience.

What it does

Our product is a mobile app that enables users to create new stories with illustrations by answering several prompts (character, setting, etc.). As the user reads through the story, the story prompts them with several paths to take. Instead of following a predefined script, our app uses generative AI to create the next part of the story based on the user's decision. Each user's stories are stored in their own library.

How we built it

We built the backend on the Node.js platform using the Express.js framework, and used MongoDB to store user data. Our server interfaces with the OpenAI API as well to generate text and images from ChatGPT and DALL-E, allowing new content to be created based on the user's interaction with the story. We stored and used previous story context and ChatGPT conversation history when generating new content in order to create cohesive storylines.

In addition, we used Microsoft Azure Blob Storage for hosting DALL-E generated images, enabling stories to be saved over the long term. The server is hosted on Microsoft Azure App Services.

To create a smooth user experience with a modern interface, we built the frontend using the React Native framework.

Challenges we ran into

With many tasks to complete and different skillsets within our team, team coordination was challenging at first, but we were able to quickly learn new technologies and become increasingly efficient over the course of the weekend.

Storybooks also contain a large amount of text and images, so some limiting factors for us were the rate limit and token limit for making API calls, as well as the speed of image generation.

Accomplishments that we're proud of

We were able to design and implement a functioning full-stack mobile app. The app has an intuitive, elegant user interface. Our product also integrates with MongoDB and Azure cloud storage to save all of a user's previous storybooks. We were also able to utilize ChatGPT and DALL-E with our interactive features to generate stories that flowed well, were relevant and specific to the user's initial prompts, and were engaging.

What we learned

Through this project, we learned how to use the OpenAI API to use ChatGPT and DALL-E for text and image generation. We also learned how to specifically design queries to ChatGPT and DALL-E to achieve more suitable outputs.

We also gained new skills in full-stack development, including designing an application, learning new frameworks, and strengthening our debugging abilities.

What's next for Aesop AI

We plan to release our mobile app on the Google Play Store and iOS App Store. Some potential features to implement include editing/backtracking on existing storylines, optimization of the timing for which multiple images or paragraphs of text are generated and served to the user to reduce delays on the frontend, and story sharing among users.

Share this project:

Updates