ClariFact AI

Protecting you from misinformation. See the truth. Instantly.

Comment

Inspiration

In an era where "fake news" and deepfakes can spread across the globe in seconds, we felt the urgent need for a unified, accessible truth-seeking tool. The inspiration for ClariFact was simple: truth shouldn't be buried behind complex tools or subscription walls. We wanted to build a "Swiss Army Knife" for verification that meets users exactly where they are whether that's browsing the web, checking a WhatsApp message, or analyzing a viral image.

What it does

ClariFact is an AI-powered ecosystem that detects misinformation across all common media types:

  • Text & Claims: Analyzes statements and extracts core claims for verification.
  • Links & URLs Checks domain reputation, SSL status, and content credibility.
  • Images: Uses OCR to extract embedded text and analyze for manipulation.
  • Videos: Transcribes speech-to-text to verify spoken claims. All inputs are distilled into a single, intuitive Truth Score (0-100) with detailed analysis and source attribution. It's accessible via a premium Web Dashboard, a real-time Browser Extension and an instant-response WhatsApp Bot.

How I built it

Building a multi-modal platform required a robust, high-performance stack:

  • The Brain: A FastAPI backend in Python serves as the intelligence hub, orchestrating AI analysis via OpenAI, text extraction through Tesseract OCR and audio transcription with Whisper.
  • The Face: A premium Next.js 16 frontend designed with Tailwind CSS and shadcn/ui provides a seamless, professional interface.
  • The Reach: Twilio Integration powers the WhatsApp bot, while a Chrome Extension (Manifest V3) provides on-the-go protection.

Challenges I ran into

One of the biggest hurdles was unifying data formats. A low-quality voice note from WhatsApp, a high-resolution screenshot, and a complex political article all needed to be normalized. We spent significant time calibrating our heuristic-based scoring system to ensure it was both accurate and explainable, ensuring the system provides "Trust" rather than just a black-box percentage.

Accomplishments that I'm proud of

I am incredibly proud of the platform synergy. The fact that the same backend engine can power a web dashboard and a WhatsApp bot simultaneously while maintaining consistent truth scores makes the system feel like a true unified ecosystem. Achieving a sub-second response time for link verification was also a major win.

What I learned

This project was a deep dive into Multi-Modal AI Pipeline Architecture. I learned how to handle asynchronous processing for heavy media files without blocking the UI. More importantly, I learned that technology is only half the battle; designing for human trust through clear source attribution is just as vital as the logic itself.

What's next for ClariFact AI

The roadmap for ClariFact is ambitious:

  • Real-time Collaboration: Community-driven verification and consensus.
  • Advanced Deepfake Detection: Moving beyond speech-to-text into frame-by-frame visual manipulation analysis.
  • Multi-Language Support: Expanding truth detection to over 50 global languages.
  • Blockchain Verification: Storing verification history on-chain for immutable truth records.

Built With

Share this project:

Updates