Speak naturally. Handless understands your screen, navigates web DOM trees, streams speculative dictation in sub-100ms, and executes native Windows hardware actions without fragile mouse coordinates.
Click any voice command pill below to simulate how Handless routes intents, inspects elements, and manipulates Windows without touching the mouse.
Every layer was designed from scratch to overcome traditional voice assistant lag and fragile coordinate clicking.
Streams discrete 200ms audio chunks in parallel background threads to Groq Whisper Large v3. Text slams onto the cursor in sub-100ms when you release the push-to-talk key.
Automatically wakes Chromium's hidden accessibility engine and bridges with Chrome DevTools Protocol (CDP) on localhost:9222 to read raw web DOM nodes, attributes, and ARIA labels.
Press Win + Shift + Space over any desktop window or code editor. Handless captures the screen and answers questions or extracts text via Gemini 2.5 Flash and Groq Vision.
Direct control over Windows master volume, display brightness, power states (sleep, lock, restart), window snapping, and night light via native COM and WMI APIs.
Inspect elements live with a glowing bounding crosshair snapper. Handless automatically generates 3-tier resilient locators: Primary AutomationId, Secondary Structure, and Tertiary Vision.
Engage in continuous, low-latency spoken conversations with Gemini Live. Features a geometric 3D thinking orb that pulses dynamically with speech and executes background tools.
See how Vyoma Handless compares against traditional speech engines and cloud desktop assistants.
| Capability / Benchmark | Vyoma Handless™ | Traditional Voice Assistants | Generic Cloud STT |
|---|---|---|---|
| Perceived Dictation Latency | < 100ms (Speculative) | 1,200ms – 2,500ms | 1,800ms+ |
| Windows Hardware Control | Direct Native WMI / COM (Volume, Brightness, Power) | Limited or Simulated Keyboard Shortcuts | None (Speech only) |
| Web Browser Automation | Chromium Accessibility + CDP Bridge | Blind Coordinates / OCR only | None |
| Screen Understanding | Multimodal Gemini 2.5 Flash + Screen Lens | None | None |
| Offline Capability | 100% Offline Whisper Base int8 (Sherpa-ONNX) | Proprietary Cloud Required | Cloud API Only |
| License & Source | 100% Free & Open-Source (MIT) | Closed / Proprietary | Paid API Tokens |
Choose between downloading the standalone portable Windows release or installing directly via Python in your development environment.
No! Handless includes a 100% offline speech recognition engine (Sherpa-ONNX int8 Whisper Base) and local regex rules that operate entirely on your CPU without any internet connection. You can optionally add a free Groq API key (for cloud Whisper Large v3) and a free Google Gemini key (for Screen Lens & Gemini Live) directly in Settings.
Handless is currently exclusive to Windows 11 and Windows 10 (64-bit). It deeply hooks into native Windows OS subsystems including Windows UIAutomation (UIA COM), Core Audio Endpoint APIs, WMI display brightness controls, and Win32 window management hooks. Because these low-level APIs do not exist on Android or iOS, Handless cannot run on mobile devices. You can read all documentation and watch demos here on your phone, then download Handless on your Windows PC!
While you hold F8, Handless buffers your live microphone stream into small 200ms audio frames. As soon as you finish a phrase, it pre-transcribes the audio in parallel background threads. When you release F8, the text is already transcribed, resulting in instantaneous sub-100ms text typing into your active application.
Handless has a hardware-level Emergency Stop built-in. Press Ctrl + Shift + Q (or tap Escape twice) at any time to instantly kill all in-flight mouse, keyboard, and window automation tasks.
Yes! Handless includes a Macro Recorder (`"start macro recording"`) to record your own keystrokes and clicks into voice macros. Developers can also easily add new patterns to `LocalRuleParser` or build native Python plugins.
Vyoma Handless is a native 64-bit Windows software package (.exe) designed for Windows 11 / 10 PCs. It cannot run or install on Android phones or tablets.