Getting Started

Up and running in under 2 minutes.

1

Download & Install

Open the DMG, drag Auris to Applications, launch from Spotlight or Applications folder.

2

Grant Permissions

Microphone
Accessibility
Input Monitoring
Screen Recording (optional)
3

Voice Model Downloads

The on-device transcription model (~200MB) downloads automatically on first launch. Takes about 30 seconds.

4

Press & Speak

Hold your hotkey (default: โŒฅ Option), speak naturally, release. Text appears wherever your cursor is.

Dictation vs AI Command

Two modes, one hotkey each.

Dictation

You talk, it types.

Hotkey: โŒฅ Option (hold)
You say: The words you want to appear
Result: Polished text pasted
Say: "hey john following up on yesterday's meeting"
Get: Hey John, following up on yesterday's meeting. Let me know if you have any questions.

AI Command

You instruct, it acts.

Hotkey: โŒ˜+โŒฅ (select text first, then hold)
You say: An instruction on what to do
Result: Transformed text, answer, or action
Select a paragraph, say: "make this shorter"
Get: Concise rewrite pasted back

When to use which?

โœ๏ธ Writing something new? โ†’ Dictation
๐Ÿ” Acting on existing text? โ†’ AI Command
๐Ÿ”„ Rewriting what you typed? โ†’ AI Command (select it, say "rewrite")

Screen-Aware AI

AI Command can see your screen. No screenshots to take, no copy-pasting errors into ChatGPT.

1

Enable it

Settings โ†’ AI โ†’ Screen-Aware AI Command โ†’ toggle on. Grant Screen Recording permission.

2

Choose capture mode

Full Window โ€” captures the focused app automatically
Select Region โ€” drag a box to capture a specific area
3

Just ask

Use AI Command as usual. Say "explain this", "what's wrong here", or "guide me through this".

โฑ๏ธ Performance note: Screen-Aware adds 10-15 seconds of latency (image processing). Turn it off when you don't need visual context.

Workspace

A persistent AI chat panel that floats above your apps. Research, iterate, and execute โ€” all in one place.

1

Open Workspace

Hold Control. With text selected โ†’ opens with that context. Without selection โ†’ fresh chat.

What you can do

Conversation

Multi-turn chat with memory. Ask follow-ups, go deeper, refine answers.

Research

Ask questions, get answers. Follow up with "show me a comparison table" or "explain step 3 in detail."

Iterate on Text

Select text โ†’ open Workspace โ†’ "make shorter", "more formal", "translate to Spanish." Keep refining.

Screen-Aware Questions

Attach a screenshot (camera icon) and ask "what's wrong here?" or "explain this diagram."

Run Actions

Open apps, search the web, run Shortcuts, adjust settings, execute scripts (with confirmation).

Interface

Title bar: Source app indicator ยท Follow (PiP follows active screen) ยท PiP mode ยท New Chat ยท History
Chat area: Context card (selected text) ยท Messages with Markdown ยท Streaming status ยท Suggestion chips
Input bar: Screenshot buttons ยท Text field (multi-line) ยท Send ยท Auto-send toggle (โšก) ยท โ†‘โ†“ input history

Keyboard Shortcuts

Control (hold)Open Workspace + dictate
EnterSend message
โ†‘ / โ†“Cycle input history
โŒ˜+NNew conversation
EscapeClose Workspace

Tips

๐Ÿ“‹ Select text first

The AI gives much better answers when it has context about what you're working on.

๐Ÿ“ธ Use screenshots

Error messages, UI bugs, diagrams โ€” attach a capture and ask about it.

๐Ÿ”„ Iterate freely

"Shorter", "more formal", "add bullets", "now in Spanish." Each builds on the last.

๐Ÿ†• New Chat for new topics

Start fresh to prevent context confusion when switching tasks.

Privacy: Conversations stored locally (SQLite). Screenshots saved locally. Delete anytime from History sidebar.

Settings

Customize everything to your workflow.

โŒจ๏ธ Shortcuts

โŒฅDictation (hold to record)
โŒ˜+โŒฅAI Command (select text first)
โŒฅ+โ‡ง+VPaste last transcript
โŒ˜+โŒฅ+RRefine selection (no voice)
โ‡ง+HotkeyRaw mode (bypass AI)

Double-tap lock: tap twice to start hands-free recording, tap again to stop.

๐Ÿค– AI Refinement

Modes:
Refined (default) Summarized Raw
Quality Tiers:
Fast~0.5sQuick replies
Balanced~1sEveryday use
Premium~2sImportant emails
Best~3sCritical writing
Tones:
Natural Professional Casual Concise Friendly

๐Ÿ“ Text & Snippets

  • Snippets โ€” voice-triggered text expansions. Say "my email" โ†’ pastes your email address.
  • Word Removals โ€” auto-strip filler words (um, uh, like, you know)
  • Word Remappings โ€” auto-correct misheard terms (e.g., "cloud watch" โ†’ "CloudWatch")
  • Always copy to clipboard โ€” keep transcript on clipboard after paste

๐ŸŽ™๏ธ Voice & Mic

  • Input device โ€” select mic from dropdown
  • Transcription model โ€” Parakeet v3 (multilingual) or Fast 110M (English, fastest)
  • Sound pack โ€” Glass (subtle) or Marimba (playful)
  • Voice calibration โ€” read a passage to measure accuracy

Per-App Intelligence

Set it once, forget it forever. Auris adapts automatically.

MailProfessionalFormal, proper salutations, no contractions
SlackCasualConversational, contractions, emoji-friendly
Jira / TicketsConciseFewest words possible, no pleasantries
Docs / NotionNaturalClean up without changing style
TerminalRawNo refinement โ€” exact transcription

Configure in Settings โ†’ AI โ†’ App Overrides. Each app can have its own tone, tier, and mode.

Tips & Tricks

๐ŸŽง Use a headset

A boom mic close to your mouth = better accuracy + you can speak quietly. Any USB or wired headset works great.

๐Ÿ”‡ Speak naturally

Don't over-enunciate. Conversational volume is perfect with a headset. The AI handles the rest.

โฑ๏ธ Pause briefly before speaking

Wait ~0.5s after pressing the hotkey. The mic needs a moment to activate.

๐Ÿ”„ Use follow-ups

After an AI Command response, use AI Command again: "make it shorter", "more formal", "add examples".

โ‡ง Raw mode shortcut

Hold Shift + your dictation hotkey to bypass all AI refinement. Pure transcription, zero latency.

๐Ÿ“‹ Refine without voice

Select text โ†’ press โŒ˜+โŒฅ+R to refine with AI. No speaking needed โ€” great for cleaning up typed text.

Troubleshooting

Text doesn't paste into the active app

Check that Accessibility permission is granted in System Settings โ†’ Privacy & Security โ†’ Accessibility. Restart Auris after granting.

Hotkey doesn't work in some apps

Ensure Input Monitoring permission is granted. Some apps with custom input handling may interfere โ€” try a different hotkey combination in Settings โ†’ Shortcuts.

Dictation misses the first word

Start speaking slightly after pressing the hotkey (~0.5 seconds). The mic needs a moment to activate. Adjust minimum key time in Settings โ†’ Shortcuts.

"Copied to clipboard" instead of pasting

Auris detected no editable text field is focused (e.g., you're on the desktop or in Finder). Your text is on the clipboard โ€” click into a text field and press โŒ˜+V.

Model download stuck

Check your internet connection. The model (~200MB) downloads on first launch. Restart Auris to retry.

Screen-Aware is slow

Expected โ€” image processing adds 10-15 seconds. Use Best or Premium tier for better visual analysis. Turn off Screen-Aware when you don't need it.

Need more help? Email support@useauris.app