Focus AI

• 

  Focus AI Logo

Inspiration

Procrastination is a challenge for everyone, especially when working online. Whitelists and blacklists are one solution; however, managing them is tedious and impractical because blacklists can't keep up with the endless number of distracting websites, and whitelists are too restrictive and disruptive to update on the go.

This inspired us to use an AI Chrome Extension to dynamically determine the context of a website and decide whether to block it if it falls into distraction categories, such as videos, games, or TV shows.

What it does

Focus AI analyzes website URLs to determine their context. It blocks sites that fall into distraction categories (like videos, games, or TV shows) and allows access to productive or necessary sites.

How we built it

We use Prompt API and Gemini Nano for on-device prompts.

• The model is initialized after Chrome loads, with a prompt that defines distracting sites (TV, games, videos).
• The URL is read whenever it changes and is sent to Gemini as a prompt to determine if it's productive or unproductive.
  
• If the site deemed unproductive, we block it.

Challenges we ran into

• Ensuring the AI model could quickly and accurately classify websites in real time.
  

Accomplishments that we're proud of

• Successfully integrating AI with a Chrome extension!!!
• Building a system that automatically blocks distracting websites based on URL context!

What we learned

• How to build Chrome extensions and use Chrome APIs.
  
• How to use Gemini Nano and the Prompt API.

What's next for Focus AI

• Use website context, like page titles, to improve classification.
  
• Improve accuracy and precision in identifying distractions.
  
• Enhance handling of long URLs, like Google and Bing search URLs, to better detect distracting searches.
  

Built With

• gemini
• javascript