Voice Agents
Speak a brief, get code, copy, a call script or structured notes.
Speak a messy brief; get a clean deliverable. Voice Agents uses the browser's SpeechRecognition API, so nothing is uploaded until you press generate.
Output types
| Output | Returns |
|---|---|
| Code | Short plan plus runnable code with file paths |
| Marketing copy | Headline, subhead, 3 body variants, CTA |
| Call script | Opener, discovery, objection handles, close |
| Structured notes | Summary, decisions, action items, open questions |
Tips
- Speak the constraint first — the model weights early context heavily.
- Use the context field for anything you don't want to say out loud.
- Browser support varies; Chrome and Edge are the most reliable.
- You can always type or paste a transcript instead of speaking.