Reply Refactor – Voice typing & rewrite

Write with your voice, anywhere. Reply Refactor turns what you say into text right where your cursor is: emails, chats, forms, tickets. And when you're done, it rewrites it better. 🔒 PRIVACY FIRST Speech recognition and rewriting run on your computer. No servers, no account, no subscription: audio and text never leave your browser. Only the models are downloaded, once, and then they stay on your device. 🎙️ DICTATE AND EDIT TOGETHER • Text appears where your cursor is, sentence by sentence. • Type while you dictate: move the cursor, fix a word, keep talking. • Spoken punctuation in Italian and English: "comma", "period", "question mark", "new line"… Spoken punctuation wins over automatic punctuation, no duplicates. • Ctrl+Z undoes as usual. ✨ REWRITE IN ONE CLICK Pick a tone and get clean text: Fix, Professional, Friendly, Concise, Formal, Casual, To English, or your own instruction. Rewrite the selection or the whole field, then Replace or Copy. 🫧 ONLY WHERE YOU WANT IT The floating bubble is off by default: you turn it on for the sites you choose, one subdomain at a time, and drag it wherever you like. Everywhere else, use the Alt+Shift+D shortcut. ⚡ FAST ON YOUR GPU With WebGPU dictation is near-instant (Whisper Large v3 Turbo). Without a GPU it also runs on the CPU, more slowly — your choice. REQUIREMENTS • Chrome 124 or later. • Recommended: a graphics card with WebGPU and about 2.5 GB free for the models. • On Linux, WebGPU must be enabled in chrome://flags: the extension guides you step by step. KNOWN LIMITS • Google Docs is not supported (it doesn't use real text fields). • Rewriting uses a small on-device model: always reread before sending. Designed and built by SKYL4R S.R.L. — https://skyl4r.ai