You say the goal.
VOYA does the clicks.

VOYA is a voice-first browser agent for people who can't easily use a mouse or keyboard. Just ask.

001The Problem

The web isn't accessible to everyone. Traditional browsing depends on clicking, scrolling, and typing.

No mouseneeded
No keyboardrequired
Justsay the goal
002The Solution

Your Voice Is Enough.

Describe what you need. Let the agent handle the interactions.

Say the Goal

Describe what you want done in plain language. No mouse required.

The Agent Plans

Claude on Amazon Bedrock turns your request into browser steps.

The Browser Acts

The extension clicks, types, and scrolls until the task is done.

003Process

How It Works

From one sentence to a finished task. The agent handles every click in between.

  1. Speak

    Type your goal in plain language. Voice input is coming soon.

  2. Understand

    The agent reads the page and plans the steps to get there.

  3. Act

    It clicks, types, and scrolls for you, checking the page after each step.

  4. Complete

    The task is done, and the side panel shows what happened.

004User Interface

Meet the VOYA side panel

VOYA

One button does everything

Press it to talk, press it again when you’re done. Or tap Control twice from any web page. It does the same thing in every state, in the same place.

How it works
Button
1
From any page
Ctrl ×2
Any time
Stop

This is the real VOYA side panel, running here in demo mode with sample content. Press its button to play an example. In Chrome, VOYA hears you and does the clicks.

005Tech Stack

Technology Ecosystem

How a sentence becomes clicks: five stages from your voice to the browser.

Your Command

01

Speak it with Alt+Shift+V, or type it in the side panel.

VoiceAlt+Shift+VTextSide panel
Runs in Chrome extension

Speech Recognition

02

The backend records the mic and Google speech recognition turns it into text.

Google Speech RecognitionSpeechRecognition (Python)Typed commands skip this stepRuns in Python backend

Command Routing

03

Picks the fastest route for each command.

  • Direct

    Instant, no model call · “scroll down”, “click 12”

  • Claude Haiku 4.5

    Quick, one-step tasks · “click the search button”

  • Claude Sonnet 4.5

    Multi-step tasks · “fill out this form”

Runs in Python backend

AI Agent

04

Plans the steps and calls browser tools: read the page, click, type, fill a form, ask you.

Strands Agents SDKAmazon BedrockClaude Haiku 4.5 · Sonnet 4.5Direct commands skip this stepRuns in Python backend

Browser Actions

05

The content script numbers every element, then clicks, types, scrolls, and navigates.

Chrome Extension · Manifest V3Content ScriptRuns in Chrome extension
Chrome extensionWebSocket · localhost:8765Python backend