Inspiration
How could we create AI accessible to digitally disadvantaged communities and increase digital literacy? That's why we developed Prompto. With AI that is simplified and a UI that is familiar, users don't need to be intimidated to intensively converse with AI. Human-AI interactions can be unserious and enjoyable, like a social media platform, while still utilizing the benefits of intelligent technology.
What it does
Prompto, an AI generative platform that creates images from a user’s prompt and allows users to swipe left to dislike or right to like a generated image, adapts to curate images more closely aligned with a user’s liking preferences. Inspired by the swipe feature of dating sites like Tinder coupled with AI image tools like Dall E, we created Prompto. You begin by entering a base description for the image you’d like to see, like “show me two cats.” Prompto then generates five images using your input. You select any image of the five closest to your desired vision, and Prompto returns user input more tailored to the image you choose to guide you in making your search more specific. For example, if you selected the top left image, Prompto would return, “a couple of cats sitting on top of a wooden floor, a photorealistic painting by Hanns Katz, international gothic, handsome, symmetrical, dynamic pose,” which you could copy and paste into your next iteration of generating an image, which is a form of simplified prompt injection.
How we built it
The front-end was built using React Native via Expo. We used Modal Stable Diffusion combined with Replicate's Clip interrogator to create a robust content generation algorithm that dynamically adapts to user preferences through the experience. We used Flask to set up Web API to connect the algorithm and the mobile view. The algorithm and front-end are working independently but the glue is still in development.
Challenges we ran into
We were challenged by caffeine crashes, difficult LLMs, dependencies, linking front-end and back-end, and artificial unintelligence (LOL).
Accomplishments that we're proud of
We're proud of pulling an all-nighter, drinking 12.2 cans of Red Bull across the team, and launching the app.
What we learned
We learned to use React Native, app development, about LLMs, and making friends.
What's next for Prompto
The future for Prompto involves successfully integrating backend and frontend technologies, user profiles, and a liking and commenting system.
Built With
- flask
- modal-stable-diffusion
- python
- react-native
- replicate
Log in or sign up for Devpost to join the conversation.