Yes, the 1000th macOS dictation app. But hold a key and clean text hits your cursor in about a second — fast as hell, and it spells your weird jargon right instead of writing "cubernetties". The twist: a second hotkey runs an agent that actually operates your Mac — sorts files, drafts replies, controls apps. Plug in any MCP server and its tools become the agent's. Can your dictation app do that? Didn't think so. EU-hosted, native, no Electron.

Hey Product Hunt. Yes, I built the 1000th macOS dictation app. Somewhere a VC is funding the 1001st as we speak. In my defense, I didn't stop at dictation. The boring part every app has: hold a key, talk, clean text appears at your cursor. Except this one is fast as hell — text lands in about a second — and it spells your weird jargon right (say "Kubernetes", "Postgres", "lisper" — it nails them instead of guessing). The part nobody else has: a second hotkey talks to an agent that actually runs your Mac. Not "here's a helpful suggestion 🙂" — it does the thing. "Tidy up my desktop" and it sorts your files. "Open the latest invoice and tell me the total" — done. It reads your screen, files and calendar, picks its own tools, and asks before it touches anything. And for the nerds: plug in any MCP server (GitHub, Notion, databases, whatever) and its tools instantly become the agent's tools. I stopped writing integrations by hand like a caveman. EU-hosted, keys stay server-side, native Swift, no Electron. 24h trial, no signup, no soul required. Roast me in the comments — that's why I'm here. AMA about the architecture or the EU-hosting tradeoffs.