Secure Docs Tagger

Identify the confidentiality level of Google Docs locally with our Chrome extension. Powered by an on-device AI model, it ensures privacy, providing clear classifications and detailed explanations

Comment

Inspiration

The inspiration for creating the "Secure Docs Tagger" extension came from the growing need to manage document confidentiality in collaborative environments, especially on platforms like Google Docs. With the increase in sensitive information being shared online, it became evident that users needed a tool that not only identified the confidentiality level of a document but also provided clear justifications for such classifications.

What it does

"Secure Docs Tagger" is a Chrome extension that allows users to identify and visualize the confidentiality level of a Google Docs document. The extension uses the Google Docs API to extract document content and leverages AI APIs integrated into Chrome to classify the content into different confidentiality levels, also providing detailed justifications for each classification.

How we built it

The extension was developed using standard web technologies such as HTML, CSS, and JavaScript. Communication with the Google Docs API was established to access document content, using OAuth for secure user authentication. Additionally, Chrome's AI APIs, specifically the Prompt API, were integrated to perform the classification.

Challenges we ran into

During development, I faced several significant challenges, including:

OAuth configuration on Google Cloud: I had no prior experience with OAuth, so setting it up on Google Cloud was a challenge.

Integration with Chrome's AI APIs: Being experimental APIs, unexpected issues arose. One major challenge was that the model would crash and stop working, requiring me to install different Chrome versions for testing.

API instability: Some APIs worked correctly in one browser but not in another, even with the same code. Moreover, as experimental APIs, there were times when they did not function properly.

Accomplishments that we're proud of

The project development had to be done at an accelerated pace, as I learned about the challenge less than a week before the deadline. This required me to quickly learn about Chrome's integrated APIs and OAuth authorization implementation in the context of Chrome extensions. It was an intense process but also an opportunity to gain new knowledge.

What we learned

During the development of "Secure Docs Tagger," I acquired new knowledge, such as configuring OAuth, a process I had not previously undertaken. I also deepened my understanding of the integrated APIs available in Google Chrome and the creation of extensions for this browser.

What's next for Secure Docs Tagger

  • Multilingual support: Incorporate the ability to classify documents in multiple languages, using Chrome's integrated APIs for language recognition and translation.

  • Model optimization: Iterate and refine the prompt to improve the accuracy and relevance of classifications.

  • Local storage: Implement functionality to store classifications directly in Chrome, avoiding reclassification of documents whose content has not changed.

  • Personalized recommendations: Provide suggestions for handling the document based on its confidentiality level, utilizing Chrome's Writing API.

  • Expansion to other services: Extend functionality to other Google products, such as Sheets, Slides, Gmail, and more, to offer a more comprehensive experience.

Share this project:

Updates