VOX
Windows voice assistant that does the clicking: say what you need in English or Hindi and VOX plans it, runs it on your computer and answers out loud.
Overview
VOX (formerly Arora AI) is a voice-first desktop agent for Windows. Press Ctrl + Space or say the wake word, ask for something in English, Hindi or a mix of both, and four agents take it from there: a Listener works out what you mean, a Strategist turns it into a plan of tool calls, an Executor runs each step, and an Overseer checks the result and retries if needed. It drives the browser, reads what is on your screen, runs several jobs in parallel, and asks a clarifying question when a request is ambiguous. Every action passes a risk engine first: risky steps wait for your approval, dangerous shell and file operations are blocked outright, and unknown tools are denied by default. A local mode keeps everything on your machine.
Key features
- Ctrl + Space or wake word, then just say it
- English, Hindi and Hinglish understood and answered
- Four-agent pipeline: Listener, Strategist, Executor, Overseer
- Browser automation: searches, compares and reads pages for you
- Reads your screen and answers about what it sees
- Runs several tasks in parallel and saves routines
- Approval before risky actions; dangerous commands blocked
- Local mode for private use
Pipeline
- 1
User presses Ctrl + Space and speaks
- 2
Listener detects language and intent
- 3
Strategist plans the tool calls
- 4
Risk engine approves, asks, or blocks each step
- 5
Executor runs it; Overseer verifies and VOX answers out loud
