I got tired of constantly typing while doing research and development work.
So I built an AI voice assistant Chrome extension that lets you control
any website by voice.
Instead of typing "search for kubernetes documentation", you just say it.
It searches, follows links, fills forms—all via natural voice commands.
How it works:
- Install the extension (one click)
- Enable microphone permission
- Say "search for [query]" or "click the first link" on any site
- It understands natural language (not just rigid commands)
Why I built this:
I noticed that voice assistants exist (Google Assistant, Alexa), but they don't
really work with websites. They can't fill out forms, navigate pages, or
interact with the DOM. So there's this gap where you need voice control,
but you still have to use your hands.
Current state: Launched 4 days ago, 0 users (hence this post ).
I've been gathering feedback and iterating based on early testers.
Technical approach:
- Built with WebSpeech API for recognition
- Uses Claude/GPT for intent understanding
- Local processing where possible (privacy-first)
- Works offline for core features
What I'd genuinely appreciate feedback on:
1. Does the permission model make sense to you? I'm asking for microphone access only.
2. What voice commands would be most useful for your workflow?
3. Any privacy concerns I should address more clearly?
4. Should I open-source the core logic? (Considering it)
jatinlalit•21h ago
I got tired of constantly typing while doing research and development work. So I built an AI voice assistant Chrome extension that lets you control any website by voice.
Instead of typing "search for kubernetes documentation", you just say it. It searches, follows links, fills forms—all via natural voice commands.
How it works: - Install the extension (one click) - Enable microphone permission - Say "search for [query]" or "click the first link" on any site - It understands natural language (not just rigid commands)
Why I built this: I noticed that voice assistants exist (Google Assistant, Alexa), but they don't really work with websites. They can't fill out forms, navigate pages, or interact with the DOM. So there's this gap where you need voice control, but you still have to use your hands.
Current state: Launched 4 days ago, 0 users (hence this post ). I've been gathering feedback and iterating based on early testers.
Technical approach: - Built with WebSpeech API for recognition - Uses Claude/GPT for intent understanding - Local processing where possible (privacy-first) - Works offline for core features
What I'd genuinely appreciate feedback on: 1. Does the permission model make sense to you? I'm asking for microphone access only. 2. What voice commands would be most useful for your workflow? 3. Any privacy concerns I should address more clearly? 4. Should I open-source the core logic? (Considering it)
It's completely free. Link: https://chromewebstore.google.com/detail/ai-voice-assistant-...
Happy to answer any questions. Thanks for checking it out!