Jachai AI — Making Fact-Checking Accessible to Everyone
Inspiration
In Bangladesh, misinformation spreads rapidly through social media, online news platforms, and messaging applications. Many people struggle to determine whether a viral claim or news headline is trustworthy. Language barriers and the lack of accessible fact-checking tools make this challenge even harder.
This inspired me to develop Jachai AI (যাচাই AI) — a bilingual, AI-powered fact-checking platform designed to make evidence-based verification more accessible to everyone.
What It Does
Jachai AI aims to help users investigate claims by searching for relevant information from multiple online sources and analyzing the available evidence.
It is designed to provide four possible verdicts:
- Supported: Available evidence supports the claim.
- Contradicted: Available evidence contradicts the claim.
- Partially Supported: Only some parts of the claim are supported.
- Insufficient Evidence: Available information is not enough to reach a reliable conclusion.
The platform prioritizes transparency by presenting supporting and contradicting evidence, source links, explanations, and verification limitations.
How I Am Building It
I am developing Jachai AI using beginner-friendly and accessible technologies:
- Python and Streamlit: Application development and user interface.
- OpenRouter: AI-powered claim and evidence analysis.
- Tavily: Online evidence and source discovery.
- GitHub: Version control and project collaboration.
- Streamlit Community Cloud: Application deployment.
The application is designed to support both Bangla and English, making fact-checking more accessible to Bangladeshi users while remaining useful internationally.
Development Approach
I am following the AI-Driven Development Life Cycle (AI-DLC):
- Understand — identify the problem and user requirements.
- Analyze — evaluate technical requirements and limitations.
- Design — plan the architecture and user experience.
- Build — implement the application step by step.
- Test — check functionality and identify errors.
- Validate — evaluate reliability, evidence quality, and security.
- Deploy — prepare the application for public access.
Challenges and Learning
As a beginner in AI application development, integrating AI services, web search, evidence analysis, and bilingual interfaces is a significant learning experience.
One of the biggest challenges is ensuring that AI-generated conclusions are grounded in actual evidence rather than unsupported model assumptions.
Through this project, I aim to improve my practical skills in AI integration, software engineering, API usage, and responsible AI development.
Our Vision
Our vision is simple:
Make reliable information easier to access, understand, and verify.
Jachai AI is not intended to replace professional fact-checkers. It is designed to help people examine claims critically, explore evidence, and make more informed decisions.
তথ্য যাচাই করুন, সত্য জানুন।
