ClearBoundaries: Preventing Harassment in Digital Communication

Inspiration

ClearBoundaries was inspired by the increasing instances of harassment in digital communication, which often go unnoticed and unaddressed. We wanted to create a tool that could empower individuals and organizations to promote respectful and professional interactions online. With the rise of online harassment, we saw an opportunity to leverage AI to automatically analyze and detect harmful language, preventing such behavior in real-time. This aligns with our mission to support gender equality and a safer, more inclusive online space for women and girls globally, especially in professional settings.

What it does

ClearBoundaries is an AI-powered solution designed to detect harassment in emails and other text-based communications. By using the Gemini API SDK, the tool analyzes text for subtle signs of harassment or negative tone, flagging potentially harmful content. Our platform enables businesses, organizations, and individuals to ensure their communication stays respectful and professional, preventing harassment before it can escalate. This helps foster safer, more inclusive environments, whether it's within a company or in personal exchanges.

How we built it

We built ClearBoundaries as a SaaS (Software as a Service) platform, focusing on offering scalable, flexible, and easy-to-access harassment detection. We chose this model over a traditional web app to reach a wider audience, allowing our solution to scale with demand and cater to diverse users from individuals to large enterprises.

We used Google’s Gemini API SDK to power the core of the harassment detection functionality. The model is already highly trained to analyze text, so we leveraged it without the need to train a custom model, ensuring accuracy and security.

Challenges we ran into

Like any ambitious project, we encountered several challenges along the way:

• Model Accuracy: While the Gemini model is robust, fine-tuning it for nuanced forms of harassment and subtle negative tones in communication proved to be challenging a bit. We continuously improved the system by analyzing real-world data and updating the prompt template to adapt to evolving patterns of harassment.

• Scalability: The ability to scale the platform for real-time analysis of potentially large volumes of data, without sacrificing speed or performance, was another key challenge. Our solution was to build a highly optimized API to ensure seamless performance.

Accomplishments that we're proud of

• Successfully built a fully functional SaaS platform that can analyze and detect harassment in real-time.
• Enabled real-time communication analysis with Gemini's powerful AI models.
• Created a tool that can help organizations and individuals maintain safe and respectful digital spaces, tackling harassment proactively.

What we learned

Throughout this journey, we gained deeper insights into scalable architecture and the importance of optimizing for high traffic volumes. This experience reinforced how AI and cloud technology can be leveraged to create powerful, scalable solutions.

What's next for ClearBoundaries

Moving forward, we aim to expand the capabilities of ClearBoundaries. We plan to enhance the AI model’s accuracy with more diverse training prompts and continue refining the platform to handle new forms of harassment. We also intend to integrate more communication platforms beyond emails, such as social media and messaging apps, phone calls thanks to Gemini for broaden the scope of our solution. Finally, we will keep optimizing the user experience to ensure that ClearBoundaries remains a reliable and efficient tool for our users.

Built With

• fastapi
• gemini
• python