pusheen.ai

Inspiration

STEM fields are often perceived as complex and challenging, presenting a high barrier to entry that can be particularly intimidating for children. Recognizing this, we developed a solution aimed at fostering enthusiasm among young learners towards STEM subjects—an AI-powered learning experience centered around everyone's favorite cartoon cat Pusheen! By blending Pusheen's playful and endearing image with educational content, we create an inviting and enjoyable learning experience that encourages curiosity and a positive attitude towards STEM from an early age.

What it does

If a child is curious about any STEM subject in particular, they are able to input a prompt into pusheen.ai, which then returns a tailored AI-generated video starring Pusheen that is crafted to explain the concepts in a manner that is not only kid-friendly, but also easy to understand. These educational videos are designed to make learning both fun and accessible, while also encouraging children's curiosity by incentivizing them to ask more questions.

How we built it

We prompted OpenAI's ChatGPT API to generate a story about a certain STEM topic that the user inputted. This story was fed into a Stable Diffusion API (specifically focused on our Pusheen Cat) to generate images from text to pair with the story. Then, the story is fed to ElevenLab's API to generate voice narration from text. Finally, utilizing Google Cloud Platform - the images and audio narration are combined to form an informative video for the user!

Challenges we ran into

Programming on a lack of sleep is hard. Also, trying to get a consistent image generated from Stable Diffusion that had little deformities or errors was a tough challenge we had to tackle.

Accomplishments that we're proud of

We are proud of not only successfully delivering a highly technical project that was interesting to build, but also taking time to create a pleasant design and user experience. This is something that we could see ourselves or younger siblings actually use!

What we learned

On the technical side, we learned quite a bit from working with Stable Diffusion (took quite a lot of playing around to get a consistent Pusheen character generated) and the ChatGPT API. We also got to work with React and Tailwind.css, which was helpful for transferring our vision on Figma to reality.

What's next for pusheen.ai

We're applying to Y-Combinator for Summer 2024! Pusheen to the moon.

Jokes aside, there are additional improvements to the site we would love to make in the future. This includes features such as account creation (to save previous prompts and videos), interactivity with the story for engagement (ex. questions), and additional animation/details for the user interface.