Dictation that never leaves your Mac.

Yes. Parlo runs speech recognition and language processing on-device, so dictation, meeting notes, and voice commands all work with Wi-Fi off after a one-time model download. Your audio never leaves the Mac. Two things send text, never audio: factual questions (“exchange rate USD→EUR”, “who's the president of Peru”) send a short text query to keyless public sources and a local model reads the answer back, and cloud meeting notes — only if you turn them on in Setup with your own API key — send the meeting transcript to the provider you chose.

Yes. Parlo's dictation removes fillers and false starts, inserts spoken punctuation, and applies spoken self-corrections automatically (“meet at 2, actually 3” → “meet at 3”) — all processed on-device, in any macOS app.

Act is Parlo's voice-command mode (currently in alpha). It handles app switching (“open Slack”, “switch to Finder”), reminders (“call the vet tomorrow at 9”), file operations (“copy report.pdf to the desktop”), and quick math or lookups (“what's 15% of 200”, “capital of Peru”). Anything destructive asks for confirmation before it runs.

Parlo detects a call is happening (calendar invite or sustained mic activity), offers to record, then captures your mic and the system audio locally — no bot joins the meeting. It also notices when the call ends and offers to stop the recording, so it doesn't keep running after you hang up. When you stop, it writes a transcript and structured notes to a folder on your Mac. By default everything stays on-device and the audio never leaves your Mac. If you want stronger notes, two opt-in options are built in: Polish opens the transcript in ChatGPT or Claude with a ready-made prompt (the transcript lands on your clipboard — paste and enter), and cloud notes generate them with OpenAI, Anthropic, or any OpenAI-compatible endpoint using your own API key. Cloud notes send only the transcript text, never the audio, and if a cloud run fails Parlo falls back to the on-device model — notes are never missing.

Yes. Parlo can publish each meeting's notes, transcript, and metadata (never the audio) to a folder you pick — your iCloud Drive, a network share, or natively to Google Drive (My Drive or a Workspace Shared Drive, as a real Google Doc or markdown). Turn it on in Setup → Meeting Sharing, share that folder with your team once in the provider's own app, and everyone in it sees every meeting you share. Auto-share (optional, off by default) publishes each meeting as soon as its notes are ready. Google Drive access is scoped to only the files Parlo creates — it can't see or touch anything else in your Drive.

Setup lets you choose between two on-device speech engines, picked by tradeoff rather than name: “More languages” (the default) covers almost any spoken language, and “Speed & accuracy” is faster and more accurate but covers 25 European languages — English, French, German, Spanish, Italian, Portuguese, Dutch, and 18 more. Pick whichever matches what you speak; you can switch anytime in Setup.

Now: dictation, meeting notes, and voice commands in one Mac app. Next: more commands, app-aware actions, and voice search across your meeting notes. Later: sharper local models tuned from real usage, dictation on iPhone, and your phone as a capture remote. Alpha users steer all of it — Parlo evolves around what they need.

Wispr Flow is cloud-based dictation — your audio is transcribed on a server. Superwhisper does dictation and meeting notes, with a local mode. Parlo is fully on-device by default and bundles three things in one app: system-wide dictation, bot-free meeting notes, and voice commands that operate your Mac — plus an opt-in cloud-notes path with your own OpenAI or Anthropic API key, so you keep control of where notes are generated. Pricing is $8/month with a 14-day free trial, or $80 lifetime for the first 100 users — versus roughly $15/month for Wispr Flow and $8.49/month or $249.99 lifetime for Superwhisper.