Frequently asked questions.

Agent Mode is a continuous voice agent. Press fn + shift and just talk. It listens, talks back, and takes action across your connected apps (like creating a Linear ticket and posting the link in Slack), searches the web, and works with files and reminders on your Mac. It confirms before it sends, creates, or changes anything.

Hold fn, speak, release, and polished text appears at your cursor in any app. It cleans up filler words and punctuation automatically. Use fn + space for hands-free dictation.

No. Dictation is included and works system-wide, but the heart of mrmr is Agent Mode: using your voice to get real work done across apps and your Mac.

Yes. Hand bigger tasks (research, summaries, multi-step lookups) to a background sub-agent and keep talking while it works. Ask for the status anytime, and every run is saved — come back later, even in a new conversation, and just describe the task to get its result.

Yes. Save your own scripts (shell, Python, Node, anything your Mac can run) or import your Raycast Script Commands, then run them by voice in Agent Mode. Speak any arguments out loud, and the script's output flows back to the agent so it can keep working with the result. Imported and installed scripts stay inactive until you've reviewed and trusted them, and every run is tracked in a Runs tab.

Yes. Agent Mode includes on-Mac tools: it can search files on your Mac (confined to your home folder) and search your browser history and bookmarks, then act on the result: open it, summarize it, or use it in the next step.

mrmr's dictation supports around 60 languages. In Voice settings you can lock recognition to a single language, bias toward a few you regularly mix, or leave it on auto-detect to switch live as you speak.

No. Both modes need an internet connection for transcription and the agent.

For getting work done in third-party apps, yes. Siri handles Apple's own apps and simple system tasks, and as of mid-2026 its cross-app abilities in third-party apps are still in testing, per Apple's own announcements. mrmr holds a continuous conversation and takes confirmed actions across Slack, Linear, Google Calendar, Gmail, Notion, and more today.

Wispr Flow and Superwhisper are dictation tools: they turn speech into text, and by their own docs they do not take actions in other apps. mrmr includes system-wide dictation too, but its core is Agent Mode: a voice agent that takes confirmed actions across Slack, Linear, Google Calendar, and more. See the full dictation app comparison.

Dottie is a free, open-source voice agent that drives Apple's built-in apps (Mail, Messages, Calendar, Notes) by controlling them through the accessibility layer, and it can run fully on-device. mrmr acts through managed OAuth connectors to third-party work apps like Slack, Linear, Google Calendar, Gmail, and Notion, plus your own scripts, and confirms every write. If your work lives in those apps, mrmr reaches them directly; if you live in Apple's own apps and want local models, Dottie is a good fit. Full comparison in our best AI assistant for Mac guide.

Yes. mrmr uses OAuth for all integrations; we never see or store your passwords. Permissions are scoped to only what's needed, and you can revoke access at any time from your account settings in each app. Your credentials never enter the AI model's context.

mrmr needs Accessibility permission (to insert text at your cursor) and Microphone access (to capture your voice). Screen Recording is optional: grant it only if you want the agent to answer questions about what is on your screen. Agent Mode also asks for Automation permission the first time it works with Apple Reminders. All standard macOS permission prompts, and nothing runs in the background without your knowledge.

The microphone only opens when you trigger mrmr; there's no always-on listening, and when Agent Mode is idle it sends no audio. Audio is processed in real time to power transcription and the agent. If Save transcripts is enabled, dictation recordings are retained locally on your Mac and can be deleted or disabled from Privacy settings; otherwise they are deleted after processing. Audio is not used for training.

mrmr is in private beta. Join the waitlist and we'll send access as spots open up. For the fastest path, book a 20-minute setup call.

mrmr is in private beta. Pricing will be announced before paid plans launch, with founder pricing for early beta users.