• 

  See how Linked Summary connects AI summaries to the original text

🧩 Linked Summary

Summaries you can verify — each linked back to its source.

🚀 Inspiration

AI summaries are everywhere, but they often lose context and trust.
We wondered — if RAG can improve trust by citing external sources, why shouldn’t we do the same for internal ones?
Linked Summary was born from that question: making on-page summaries verifiable by linking them back to their origins.

💡 What it does

Linked Summary is a Chrome extension that uses Chrome’s built-in AI to summarize any webpage or article.
Unlike other summarizers, its key feature is that it connects every sentence of the summary directly to its original source in the text.

This allows users to:

• Generate concise summaries directly in the browser
• Click any summary sentence to instantly scroll to its corresponding paragraph or sentence
• Verify the accuracy and context of AI-generated summary in one click
• Keep all data local, ensuring privacy and speed — no cloud calls, no tracking

🛠 How we built it

We used the Chrome built-in AI API for on-device summarization, then applied transformer-based semantic embeddings (@xenova/transformers) to find semantically close relationships between summaries, paragraphs, and sentences.
The frontend, built with lit, provides an efficient, lightweight interface for exploring summaries and sources side by side.

⚙️ Challenges we ran into

The main challenge was finding semantically closest relationships between summaries, paragraphs, and sentences.
We needed to capture meaning beyond surface-level similarity, linking each AI-generated phrase to its true origin within complex, dynamic HTML.
Optimizing this semantic matching for both accuracy and speed — while running entirely in-browser — became the core technical challenge of the project.

🏆 Accomplishments we’re proud of

• Built a fully local summarization pipeline with Chrome’s built-in AI — no external API calls
• Designed a semantic anchoring algorithm that connects summaries to exact DOM segments
• Delivered a fast, transparent, and privacy-first UX that makes AI output verifiable

📚 What we learned

• How to integrate on-device AI for real-world use inside Chrome extensions
• Building robust semantic alignment systems that survive dynamic DOM updates
• The importance of explainability and user trust in every AI interaction

🔮 What's next for Linked Summary

• Extend coverage to dynamic and script-heavy webpages
• Add interactive highlight effects and smoother navigation

Built With

• javascript
• transformers
• webassembly