Inspiration

For us and many of our fellow students, reading dense blocks of text from textbooks can be a real struggle. They feel overcomplicated, long, and confusing. They aren't good for visual or auditory learners and don't provoke any engagement with the content, which would greatly improve recall. SmartCookie incorporates all of these elements and simplifies the text so that you can attack learning from all angles and get the most from your textbook.

What it does

SmartCookie is a Chrome extension. When you use it and highlight a dense block of text, it fetches matching YouTube videos and lets you choose from them. It also has a notes feature that generates succinct bullet point notes that you can save for later. To elevate engagement and improve understanding, SmartCookie also has a quiz feature that generates a 3-8 question quiz directly based on the highlighted text.

How we built it

Chrome extension's content.js listens for text highlights. When it senses that, it asks a background script to send the highlighted text to Claude Haiku, which transforms it into a 5-word YouTube search query. Using YouTube API we grab the top video results, their view counts, and their subscriber accounts. We score each video with this data, and add bias towards certain trusted educational channels (Khan Academy, Crash Course, etc.). It returns the videos and allows the user to watch them. The quiz and the notes functionality both also send the highlighted text to Claude for a response and displays it. Claude Code used in VSCode was very helpful in our production process.

Challenges we ran into

This was our first ever hackathon, and so we went in with very little idea of where to begin. Not only did we not have an idea of what kind of projects were expected, but we were also unfamiliar with the APIs and frameworks we needed to use to make our ideas work. This led us to pivot about 1/4th of the way through the hackathon to a new idea, since we felt our old one lacked the spark that we wanted. In the end it was the right choice, and we are happy with our final product.

Accomplishments that we're proud of

Going in, we really wanted to focus on an idea that was genuinely usable and applicable in our daily lives. Our initial idea was a form filler that we didn't feel like we would ever use. SmartCookie, however, is exactly the kind of thing we think would be useful for us and other students. We think it hits the perfect balance of efficiency and education; it streamlines the learning process while not encouraging students to cut corners. We are really proud of how the idea evolved from a simple video recommender to an expandable education tool.

What we learned

Firstly, we learned a lot about the hackathon process, including the initial ideation and expanding upon it as we go. Throughout this process we also learned a lot about linking APIs to serve a larger purpose than they can each do alone. We learned to treat APIs like black boxes that we give an input and an output, and our job was just to code the pipes between them.

What's next for SmartCookie

Some features are curated to make a smooth demoing process. For example, upon highlighting text, SmartCookie opens immediately. We don't feel this is convenient for a real use case, as it can be annoying and get in the way. In the future we would like to cater SmartCookie to be comfortable and easy to use for real students.

Built With

  • anthropicclaudeapi
  • chrome.runtime
  • chrome.storage.local
  • css
  • fastapi
  • html
  • javascript
  • pydantic
  • python
  • uvicorn
  • youtube-transcript-api
  • youtubedataapiv3
Share this project:

Updates