Inspiration

The inspiration behind BartaGuard came from the growing number of digital scams targeting everyday consumers in Bangladesh. Fake bKash and Nagad notifications, fraudulent part time job offers and courier payment scams have become familiar problems. I wanted to build something that could help ordinary people understand suspicious messages before they lose money or personal information.

How I Built It

I developed BartaGuard as a solo project using Next.js, TypeScript and Google's Gemini API. It supports text and screenshot analysis, allowing users to submit suspicious messages in Bangla, English or Banglish. Gemini analyzes the content and generates structured reports with risk levels, detected warning signs, evidence, practical recommendations and easy to understand Bangla explanations.

I also implemented interactive sample scenarios and designed the application to process submitted content without permanently storing messages or screenshots.

Challenges and Learning

One of the biggest challenges was making AI analysis useful and understandable for non technical users while avoiding misleading security claims. Supporting Bangla and Banglish, extracting information from screenshots and maintaining consistent structured AI responses required careful prompt design and validation.

Through this project, I gained practical experience in integrating multimodal AI, designing API workflows, handling user inputs and building a consumer focused application. More importantly, I learned that AI products should not simply identify problems but should help people understand them and make safer decisions.

BartaGuard's mission is simple: বোঝার আগে বিশ্বাস নয়।

Built With

Share this project:

Updates

Submission history