DreamCatcher

Inspiration

We were inspired by the AWS Generative AI workshop during the first day of DubHacks. The workshop went over Generative AI and what it could be used for. We thought the idea of making a story for your child was interesting, which led a spark to our idea of making a story out of dreams.

What it does

The app asks users every morning about their dream(s) last night. Users will describe the dream as best as they can and submit it. They will also submit an emoji expressing how they are feeling after that dream. The AI will use this to create an analysis of your dream (symbolism, psychoanalysis, and possible influences of the dream). In addition, it will create other fun things: an AI image depicting your dream, and an AI-generated story expanding on your dream's idea. You can save your dream and all of the AI produced content into your dream album for viewing in the future.

How we built it

We used AWS PartyRock to build the prototype of our application, and engineer the AI prompts. We then use a plugin called GenStack to convert our PartyRock app into a full stack Svelte (JavaScript framework) application. After that, we connected our full stack app to our AWS Bedrock instance so that the generative AI will work. Then, we improved the UI produced by GenStack to develop a beautiful UI for our app. We also added a new type of prompt for selecting a mood emoji, which PartyRock didn't initially have, as well as a system for saving and reviewing data produced by the AI persistently.

Challenges we ran into

The biggest challenge we ran into was authenticating and configuring our AWS Bedrock API access. We spent several hours trying to clear up the bugs in setting this up in our Svelte app, and also had to work through some other bugs regarding how the models were used.

Accomplishments that we're proud of

We are proud of the UI we created, which we think is highly aesthetic and puts AI generated content in a customizable web app format that current AI tools don't embody yet.

What we learned

Generative AI is becoming more and more powerful. Most of the hard work in producing this app was done by the AI itself. That gave us more time to focus on creating a better UI that attracts users, engineer our prompts, and add more useful features to our product.

What's next for DreamCatcher

In the future, we can consider adding more cool features to the app:

Social feature - users can share their dreams with their friends
Analytics feature - users can see their overall trends in their dreams, and also observe similar patterns in dreams based on geography/location/other data points
Virtual reality feature - users can rewatch their dreams (or AI-improved dreams) using AR/VR using an AI generated video of the dream based on their description