Veritas AI

Inspiration

In today’s world, misinformation spreads rapidly online. I wanted to create a tool that helps people verify claims quickly and reliably, combining the power of Generative AI with real-time research. The idea came from my personal frustration when trying to fact-check health, technology, or news claims manually.

What it does

Through this project, I learned how to:

• Integrate LLMs (OpenAI GPT, Gemini) with live search APIs (Tavily)
  
• Build a React + Tailwind frontend with smooth animations (Framer Motion)
  
• Implement backend research planning and Truth Score computation using FastAPI
  
• Manage a Vector Database (Qdrant) for retrieval-augmented generation (RAG)
  
• Handle secrets safely and prepare a hackathon-ready repository
  

How we built it

1. Frontend: React + Vite + Tailwind CSS, with animations for query processing, scoring, and source highlighting
   
2. Backend: FastAPI server that receives queries, generates a research plan, fetches sources via APIs, computes a Truth Score, and returns structured results
   
3. Research Agent: AI autonomously analyzes each claim, generates step-by-step research plans, and ranks sources by credibility
   
4. Database: Qdrant used for caching and semantic retrieval for future queries
   
5. Demo Flow: Intro animation → Question input → Research visualization → Truth Score output → Summary & plan
   

Challenges we ran into

• Integrating real-time search with AI reasoning while keeping the system responsive
  
• Designing a modern, visually appealing frontend that conveys research steps clearly
  
• Ensuring API keys and sensitive data were not exposed in the repo
  
• Making the project fully functional as a solo developer, while maintaining hackathon-ready polish
  

Accomplishments that we're proud of

• Built a fully functional AI Truth Engine that analyzes claims and generates step-by-step research plans.
  
• Integrated Live Search APIs (Tavily) and Vector DBQdrant)** for real-time, retrieval-augmented research.
  
• Developed a premium React + Tailwind frontend with smooth Framer Motion animations, making AI research visually engaging.
  
• Designed a Truth Score system that evaluates sources by credibility and presents clear, actionable insights.
  
• Achieved all of this as a solo developer, delivering a hackathon-ready, polished project.
  

What we learned

• How to combine Generative AI with real-world data for autonomous research and verification.
  
• Techniques for modern web development: React, Tailwind CSS, Framer Motion, and responsive design.
  
• Best practices for securely managing API keys and sensitive data in a GitHub repo.
  
• How to create a full-stack AI product from scratch: backend planning, AI reasoning, frontend interaction, and database integration.
  
• The importance of UX/UI in conveying complex AI workflows to non-technical users.
  

What's next for Veritas AI

• Expand multi-language support to verify claims in multiple languages and regions.
  
• Integrate additional APIs for broader real-time research (scientific publications, news outlets, social data).
  
• Enhance the AI Research Agent to automatically summarize findings and provide citations.
  
• Add user accounts and history tracking for personalized verification workflows.
  
• Explore mobile-friendly versions and browser extensions to make AI-powered fact-checking instantly accessible.
  

Built With

• javascript
• openai
• python
• react
• tavilyapi