overview
What VoiceOS does
VoiceOS has two modes. Dictation Mode types what you say into any text field: press fn, speak, and it removes filler words, fixes grammar and spelling, learns names and acronyms you use, and can format spoken items as a list; Edit Mode rewrites text you have selected. VoiceOS says it handles over 100 languages without switching. Agent Mode goes further: you ask for an outcome, such as replying to a message, scheduling an event or finding a file, and it acts through apps you have enabled, including Gmail, Slack, Notion, Linear, Google Calendar, Google Drive, Google Docs, Finder, Reminders, Messages and Obsidian. It can use screen context you point at, and actions that send, book or change something pause to show what they will do first.
You install the desktop app (macOS or Windows), grant permissions and connect services in Settings; developers can add their own tools with MCP servers. VoiceOS uses cloud processing, so it is not an offline dictation app. It says transcripts are stored on your device, and with Help Improve VoiceOS off, raw audio and transcripts are deleted right after processing and not used for training. Mobile apps are listed as coming soon. VoiceOS is made by WakoAI, Inc.
