Cognify

Cognify

Inspiration

We live in a time when AI can generate convincing misinformation faster than humans can verify it. Cognify was inspired by the need to restore epistemic trust — helping users know not just what content says, but how true it is.

What it does

Cognify is a Chrome extension that uses Chrome’s built-in AI APIs to analyze any webpage, article, or image and produce a real-time reliability score. It summarizes, classifies, and verifies online content against academic sources, highlighting bias or low factual quality through a simple color-coded system (🟢 reliable, 🟡 read critically, 🔴 not reliable).

How we built it

Frontend: Chrome Extension (Manifest V3, TypeScript, Tailwind)
AI Core: Gemini Nano + Chrome Prompt, Summarizer, and Proofreader APIs
Backend: Node.js + Express + Semantic Scholar API for citation matching
Caching: Redis layer for performance optimization and reduced API cost
UI: Floating overlay with real-time scores and academic references

Challenges we ran into

Coordinating multiple Chrome AI APIs in one extension context
Maintaining speed and privacy during on-device inference
Creating multilingual NLP pipelines (English, Spanish, Catalan)
Designing interpretable reliability scores that users can trust

Accomplishments that we're proud of

Achieved fully functional real-time epistemic analysis on-device
Built a modular architecture ready for open-source release
Developed an interpretable reliability model combining AI and academic data
Created a visual UX that makes critical thinking intuitive for all users

What we learned

We learned the potential of on-device AI for privacy-first browsing and how small models like Gemini Nano can enable sophisticated reasoning locally. We also discovered that epistemic analysis requires both linguistic understanding and source credibility mapping — blending technical and philosophical design.