The problem reacheck solves

In today’s world flooded with AI generated content like fake headlines, deepfake images, synthetic quotes and manipulated reviews, most people can’t tell what’s real anymore.

ReaCheck solves this by training users to detect AI generated content through a gamified, interactive platform. It:
• Boosts AI literacy in a fun, non technical way
• Helps students, parents, educators and professionals develop critical thinking around digital content
• Makes people more aware and cautious about misinformation and manipulated media
• Tailors content difficulty based on your accuracy, making it engaging for all age groups
• Can scale into educational tools, enterprise training or even browser extensions

Technologies Used:
• Next.js – for building a fast, scalable frontend
• TypeScript – strict types
• Tailwind CSS – clean, responsive UI styling
• LLMs (like GPT/Open Source) – for generating believable fake content
• OmniDimension Voice Widget – voice based interaction support
• Lightweight NLP Models – for difficulty scaling and scoring
• GitHub – version control & collaboration

By turning detection into a game, ReaCheck makes a serious problem feel engaging and empowering !

Powered by ReaCheck logic and OmniDimension magic !

Challenges I ran into

1. LLM generated content was too real.
   Our AI generated samples were so convincing even we couldn’t always tell the difference. We had to carefully balance realism and obviousness for beginner levels.
2. Smart difficulty scaling
   Building an adaptive system using lightweight NLP to adjust challenge difficulty in real time was tricky. We simplified this by using pattern based scoring for now but plan to go deeper later.
3. OmniDimension Voice Integration
   We struggled embedding the voice widget seamlessly in Next.js layout. After multiple widget placement bugs and hydration mismatches, we solved it by moving the script into the main layout file and testing across browsers.