WebMentor

Transform any webpage into an AI learning assistant, summarize, simplify, translate, rewrite, and proofread in seconds.

Comment

🌟 Inspiration

The internet is full of knowledge, but it is not always easy to understand.
As a student and lifelong learner, I often found myself jumping between tabs to summarize, simplify, translate, or proofread text just to make sense of complex webpages. That constant switching broke my learning flow.

That’s where WebMentor was born, to make learning on the web smoother, simpler, and more accessible for everyone, no matter their language or background.

🧠 What it does

WebMentor turns any webpage into a personal learning coach.
With just a highlight or click, users can:

🪄 Summarize long or complex paragraphs
🧩 Simplify text for Kids, Students, or Professionals
🌍 Translate into multiple languages instantly
✍️ Proofread grammar and clarity
🔁 Rewrite for better tone or structure
🔊 Read Aloud for auditory learners

Everything runs locally using Chrome’s built-in AI, so responses are fast, private, and even work offline.

🛠️ How we built it

Framework: React + Vite (for a fast and modular setup)
UI: Tailwind CSS, Framer Motion, and Lucide Icons
Platform: Chrome Extension (Manifest V3)

APIs Used:

  • 🧩 Summarizer API – for condensing complex text
  • ✍️ Proofreader API – to fix grammar and clarity
  • 🌐 Translator API – for multilingual support
  • 🖊️ Rewriter API – to improve phrasing and tone
  • 💭 Prompt API – for dynamic local AI prompts
  • 🔠 Language Detector – to identify input languages automatically

All AI features run entirely on-device, powered by Gemini Nano through Chrome’s built-in AI capabilities.

⚡ Challenges we ran into

  • Setting up the Origin Trial tokens and getting Rewriter + Proofreader APIs to work locally took a lot of debugging.
  • Chrome’s AI model download process sometimes stalled, which made testing slower.
  • Designing an interface that floats on any webpage and still feels lightweight required fine-tuning with Framer Motion and React state management.

Each challenge pushed me to learn how to debug new, experimental APIs — and how to build smooth experiences with local AI.

🏆 Accomplishments that we're proud of

  • Built a fully local AI-powered learning extension — no servers, no latency, no privacy risk.
  • Designed an elegant, minimal interface that integrates seamlessly with any webpage.
  • Created an experience that’s genuinely useful for students, readers, and professionals worldwide.
  • Seeing WebMentor summarize and simplify pages in real time felt like pure magic.

💡 What we learned

  • How Chrome’s built-in AI works behind the scenes — from on-device models to Origin Trials.
  • The power of client-side AI for privacy-first applications.
  • How to create a user flow that blends accessibility and education seamlessly.
  • Debugging and optimizing Manifest V3 extensions with local storage and environment variables.

🚀 What’s next for WebMentor

  • Add speech-to-text so users can talk to WebMentor.
  • Introduce on-page learning mode with interactive explanations and quiz hints.
  • Expand support to Firefox and Edge using hybrid AI models.
  • Publish on the Chrome Web Store to make it available for everyone.

✨ Built for

Google Chrome Built-in AI Challenge 2025

🔒 Privacy First. AI Smart. Learning Made Simple.
👩‍💻 Developer: Ayesha Dawodi

Built With

  • chrome-built-in-ai-api
  • framer-motion
  • language-detector
  • lucide-icons
  • manifest-v3
  • prompt-api
  • proofreader
  • react
  • rewriter
  • summarizer
  • tailwind-css
  • translator
  • vite
Share this project:

Updates