TruthTracer

TruthTracer is an AI-powered tool that verifies claims in real-time, cutting through misinformation with sourced evidence and clear, trustworthy analysis.

Comment

Inspiration

In an era flooded with misinformation, we wanted to build something that goes beyond basic fact-checking — a trustworthy AI agent that not only detects falsehoods but provides transparent, sourced, and contextual insight. TruthTracer was born to empower journalists, researchers, and everyday readers to verify claims instantly and confidently.

What it does

TruthTracer is an AI-powered fact-checking assistant that evaluates any block of text — from articles to social media posts — and verifies the factual accuracy of each claim in real time. It assigns truth ratings (True, False, Misleading, Unverifiable), explains each result with supporting citations, and presents a structured, exportable report.

It doesn’t just check — it traces the origin of the claim, highlights potential bias, and flags time-sensitive or outdated data.

How we built it

  • Frontend: Built with Next.js, deployed on Netlify, designed for minimalism and clarity.

  • Backend: Developed using NestJS with TypeScript, following clean code architecture. Hosted on AWS EC2, secured and routed via Nginx reverse proxy.

  • AI Core: Integrated with Perplexity’s Sonar, Sonar Deep Search and Sonar Reasoning Pro API for real-time reasoning, citations, and follow-ups.

  • Database: A PostgreSQL instance is prepared and partially integrated for storing user submissions, claim histories, and ratings (full use planned in the next phase).

Challenges we ran into

  • Ensuring AI evaluations are context-aware without introducing hallucinations.

  • Managing response time while using external APIs for deep research.

  • Designing a trustworthy UX that remains simple without hiding critical nuance.

  • Building infrastructure while maintaining modularity across services.

Accomplishments that we're proud of

  • Built and deployed a production-ready, full-stack AI verification system.

  • Created an extensible, clean architecture backend, setting the stage for scale.

  • Designed a sleek and mature UI that communicates authority and simplicity.

  • Integrated real-time claim validation with verifiable citations and reasoning chain — not just raw LLM output.

What we learned

  • True claim validation at scale needs more than AI — it needs systematic structure, clean UX, and evidence-first thinking.

  • Sonar APIs, especially Sonar Pro, are powerful when paired with a thoughtful prompt and structured reasoning chain.

  • Clean architecture allows new features like database storage, browser extensions, and bias detection to be layered in seamlessly.

What's next for TruthTracer

  • 📘 Full PostgreSQL integration for persistent user history, analytics, and claim audits.

  • 🧠 Upgrade to Sonar Reasoning Pro and Sonar Deep Search for deeper multi-hop verification and claim clustering.

  • 🧩 Add input options for URLs, PDFs, or pasted social media articles.

  • 🔍 Browser extension to fact-check articles in context.

  • 🏢 Launch a subscription-based dashboard for research teams, journalists, and legal firms.

  • 💼 API access for developers to embed verification into their own platforms.

Share this project:

Updates