Nav Voice

Inspiration

Navigating the web can be a frustrating and overwhelming experience for users with low vision. Screen readers exist, but they can be robotic, complex to navigate, or rely on cloud APIs that compromise user privacy. For this hackathon, I built a solution: an intelligent, privacy-first Chrome extension that 'speaks aloud' the context of any webpage.

What it does

My extension doesn't just read HTML tags; it understands the page. This is achieved by piping the page's context into Gemma, running entirely locally via Ollama. This means users get intelligent, conversational audio summaries and navigation cues with zero latency and zero data leaving their machine.

How we built it

This is built using the plugin provided by Claude

Challenges we ran into

Debugging the javascript running on the browser was a challenge but Claude code patiently kept giving suggestions and egged me on

Accomplishments that we're proud of

Building something useful in 2-3 hours and seeing it work

What we learned

The many features of Claude Code. I will be sure to add this to my toolbelt

What's next for Nav Voice

Extend this to mobile apps on iOS and Android

Built With

• chrome
• gemma-3
• javascript
• ollama