← All tutorials
On a Mac? See the dedicated macOS install guide.
AppleLogo Silicon vs Intel DMG, bypassing the security warning, and granting Mac permissions.

The Complete Guide to the AissenceAI Desktop App

Everything you need to install, configure, and run the AissenceAI desktop app for live interviews — from OS permissions and the privacy controls to the audio capture toolbar, stealth mode, keyboard shortcuts, and troubleshooting. Each step shows you exactly what the screen will look like.

Windows · macOS · LinuxPrivacy-first by default~5 minutes to set up
Get the installerStart the walkthrough
On this page
  1. Install the app
  2. Grant OS permissions
  3. Meet the capture toolbar
  4. VAD vs Continuous mode
  5. Stealth mode
  6. Privacy mode
  7. Resize to fit
  8. Keyboard shortcuts
  9. Run your first session
  10. Troubleshooting
  11. Updates
  12. Toolbar icon reference
  13. Settings panel reference
  14. Full keyboard shortcut reference
  15. FAQ
1
Step 1

Install the AissenceAI Desktop App

Sign in to your dashboard, open the Desktop App page, and download the installer for your operating system. The app is a native build (~80–120 MB) and installs to your user directory — no admin password needed on most systems.

Windows 10 / 11
.msi or setup.exe

If SmartScreen warns you, choose "More info → Run anyway". Publisher is AissenceAI.

macOS 12+
.dmg

Drag AissenceAI into Applications. First launch: right-click → Open → Open to bypass Gatekeeper.

Linux
.AppImage or .deb

chmod +x the AppImage, or install the .deb on Debian/Ubuntu.

Sign in with the same account you use on the web dashboard. Your subscription, prompts, and settings sync automatically — there is no separate desktop license.
2
Step 2

Grant the right OS permissions

The desktop app needs a small set of operating-system permissions to capture audio and run reliably. Granting them once is enough.

  • Microphone — required so the AI can transcribe your voice. macOS prompts the first time you start a session; Windows usually grants it silently. The mic is non-exclusive — Zoom, Teams, and Meet retain full mic access at all times.
  • System audio — required so the AI can hear the interviewer through Zoom, Meet, Teams, etc. Uses WASAPI loopback (Windows) or a macOS audio tap — a passive, read-only capture that does not interfere with any call audio.
  • Screen recording (macOS only) — required for system-audio capture on macOS Sonoma and later. Open System Settings → Privacy & Security → Screen Recording, enable AissenceAI, and relaunch the app.
  • Accessibility (optional, macOS) — only needed if you want global hotkeys to work while another app is focused.
System Settings — Privacy & Security
Screen Recording
AAissenceAI
ZZoom
MMicrosoft Teams
Click the lock to make changes
3
Step 3

Meet the capture toolbar

When you open the desktop app you get a slim, draggable toolbar. Three iconographic controls do all the work — each one means exactly one thing.

aissence — capture
mic
system
snap

The toolbar in its idle state — ready to capture.

  • Microphone icon — captures your voice from your mic. Click to start/stop. A red dot under the mic means it is actively recording.
  • Headphones icon — captures system audio (the interviewer's voice from your call). The icon is stable; a coloured dot under it tells you the state.
  • Camera/screenshot icon — snaps a coding question, whiteboard, or document and sends it to the AI for analysis.
  • A separate alert badge in the top-right corner appears only when a permission is missing — so a permission warning never gets confused with a capture state.
What each status dot under the headphones icon means
Idle
Ready, not capturing
Initializing
Asking the OS
Capturing
Recording... Xs / ~20s shown in toolbar
Transcribing
Speech being turned into text
Error
Check permissions
4
Step 4

Choose VAD or Continuous mode

In Settings → Audio you can switch between two transcription modes. Pick the one that matches your interview style.

VAD (Voice Activity Detection)

Default

The AI only sends audio to the transcriber when it detects speech. Lower cost, lower latency, ideal for back-and-forth conversations.

Continuous

Higher accuracy

Streams audio non-stop. Use this for noisy environments, soft-spoken interviewers, or when you want the AI to capture every breath of context.

You can flip between the two from the Settings panel without restarting a session.
5
Step 5

Turn on Stealth Mode for live interviews

Stealth Mode makes the AissenceAI window invisible to screen-share APIs. Zoom, Teams, Meet, Webex, and OBS will not capture it — even if the interviewer asks you to share your full screen.

  • Toggle stealth from Settings → Stealth Mode, or use the global hotkey (default Ctrl/Cmd + Shift + H) to hide and re-show the window instantly.
  • Position the window near your webcam so glancing at suggestions looks like you are maintaining eye contact.
  • Always do one practice call with a friend before a real interview to confirm the overlay is invisible on their end.
You
Tell them about Project Atlas
AissenceAI overlay visible
Interviewer
AissenceAI overlay hidden from screen share
6
Step 6

Privacy mode — off by default

The desktop app is private-by-default. Conversations stay on your device unless you explicitly opt in to syncing.

  • Open Settings → Privacy. The “Store conversation history” toggle is OFF by default.
  • When OFF: transcripts and AI responses live only in this app session and are wiped when you close it. Nothing reaches the AissenceAI servers.
  • When ON: every chat is saved to your dashboard so you can review it later from any device.
  • You can flip this toggle any time. It only affects sessions started after the change.
aissence — Settings · Privacy
Store conversation history
When off, transcripts are wiped on app close
Send anonymous usage telemetry
Helps us prioritize what to build next
Auto-delete screenshots after 24h
Recommended for shared machines
7
Step 7

Resize the window to fit your setup

The desktop window is fully resizable so you can shrink it to a thin strip or expand it for richer responses.

  • Drag any edge or corner to resize between 420×360 (compact) and 1600×1400 (full-size).
  • A compact width is ideal when stealth mode is on and you want a slim toolbar near your webcam.
  • A wider window is better for coding interviews where the AI needs to display longer code snippets.
420 × 360
900 × 700
1600 × 1400
8
Step 8

Useful keyboard shortcuts

Every shortcut works globally — even when another app is focused.

Hide / show window (Stealth)Ctrl/Cmd+Shift+H
Toggle microphone captureCtrl/Cmd+Shift+M
Toggle system-audio captureCtrl/Cmd+Shift+S
Take a screenshot for the AICtrl/Cmd+Shift+P
Dismiss the current AI suggestionEsc
9
Step 9

Run your first live session

A typical live-interview flow takes under a minute to set up. Follow these six clicks:

  1. aOpen the AissenceAI desktop app and sign in. Confirm the headphones-icon dot is grey (idle, ready).
  2. bClick the headphones icon to start system-audio capture. The dot turns blue when audio is flowing.
  3. cClick the microphone icon to start capturing your voice.
  4. dJoin your Zoom / Teams / Meet call as normal.
  5. eWatch the suggestion panel — once the interviewer finishes speaking, AissenceAI processes the clip and streams a response. A Recording… Xs / ~20s progress bar in the toolbar shows the current clip length. Expect an AI response within ~20–30 seconds of the clip completing.
  6. fWhen the interview ends, click the close (X) button. If Privacy Mode is OFF, the conversation is wiped immediately.
aissence — capture
mic
system
snap
Step a — idle, ready
aissence — capture
mic
system
snap
Step b — system audio flowing
aissence — capture
mic
system
snap
Step c-e — live AI assistance
10
Step 10

Troubleshooting

The vast majority of desktop issues fall into four buckets — fix them in this order.

No system audio?
On macOS, re-check Screen Recording permission and relaunch the app. On Windows, make sure the call audio is going through your default playback device — not a USB headset Windows treats as separate.
Microphone silent?
In Settings → Audio, pick the correct input device explicitly. Do not rely on "Default" if you have multiple mics.
AI says "Subscription required" after upgrading?
Click the "I've upgraded" button in the upgrade banner — it forces a fresh subscription check. Or simply close and reopen the app.
You see "Active paid subscription required" mid-interview even though your plan is paid?
This is almost always a transient network blip while AissenceAI is re-validating your subscription. The app now keeps you on your last-known-good plan when the check fails, so retrying the same prompt usually succeeds. If the error persists, restart the app — your subscription state will refresh on launch. If it still happens, your plan may have lapsed; check Billing in the web dashboard.
"Open Sound Settings" button does nothing on Windows?
This was caused by the app trying to launch ms-settings:sound as if it were a binary. It is fixed in the current build — the app now invokes the URI through the Windows shell ("cmd /c start ms-settings:sound") and falls back to the legacy mmsys.cpl control panel if the modern Settings app is unavailable. If you are on an older build, manually open Settings → System → Sound and pick the playback device your call uses as Default.
App will not launch on first run?
Right-click → Run as administrator (Windows) or Right-click → Open (macOS) once. Subsequent launches work normally.
11
Step 11

Updates

The desktop app checks for updates on every launch.

  • When an update is available you will see an “Update available” indicator in the toolbar — click it to download and install in the background.
  • You can also manually check by visiting the Desktop App page in your dashboard.
  • Release notes for every version are published on the Changelog page.
Reference

Toolbar icon reference

Every clickable element on the floating toolbar, what it does, and the state colors / variations you'll see. Built straight from the desktop app's source — what you read here is what you see.

Headphones — System audio capture

Ctrl/Cmd+Shift+M
Default state: idlemacOS needs Screen Recording permission
Captures the audio your system is playing — i.e. your interviewer's voice from Zoom / Teams / Meet. The icon does not change shape; the dot underneath tells you what state it is in.
Status indicator under the icon
Idle
Ready to capture, no audio flowing yet
Initializing
Asking the OS for the audio stream
Capturing
Audio is being recorded
Transcribing
Audio is being turned into text
Error
Permission missing or audio device problem
  • On macOS, if the dot stays grey: open System Settings → Privacy & Security → Screen & System Audio Recording and enable AissenceAI. Restart the app once.
  • On Windows, if you only hear silence: Settings → System → Sound, set the speaker your call is using as the Default device.
  • Click the headphones again to stop capture cleanly. Closing the app also stops capture.

Microphone — Your voice capture

Ctrl/Cmd+Shift+A
Default state: idleUses your default OS microphone unless overridden in Settings
Captures your voice from the selected input device. When recording, a small red dot appears under the icon. Used so the AI knows what you said and can build context across the conversation.
Status indicator under the icon
Idle
Microphone is closed
Recording
Voice is being captured
  • Pick the exact mic in Settings → Audio. Do not rely on "Default" if you have multiple inputs (USB headset + laptop mic).
  • In VAD mode the mic auto-segments on silence — you do not need to press the icon between sentences.
  • In Continuous mode the mic stays open until you toggle it off.

Screenshot — Whole-window capture

Ctrl/Cmd+Shift+S
Mode: ScreenshotActive when "enabled" is on in Settings → Screenshot
Captures the full visible content behind your screen and either (a) sends it straight to the AI with your auto-prompt, or (b) attaches it to your input box for you to type a custom question. Behavior depends on the screenshot mode you choose.
  • Auto mode: the screenshot is sent to the AI immediately using the prompt set in Settings → Screenshot → "Auto Prompt" (default: "Analyze this screenshot and provide insights").
  • Manual mode: the screenshot is attached to the input box. Type your question, then press Enter.
  • You can attach up to MAX_FILES screenshots before submitting — see the count in the icon's tooltip.

Selection — Snip a region

Ctrl/Cmd+Shift+S
Mode: SelectionActive when "enabled" is off in Settings → Screenshot
Same button slot as the full-screenshot icon — it just changes shape based on which capture mode you have active. In Selection mode you draw a rectangle around the part of the screen you want, and only that region is sent to the AI.
  • Use Selection for coding interviews where the page has a lot of UI chrome you do not need to send.
  • Esc cancels an in-progress selection.
  • You can flip between Screenshot and Selection mode in Settings → Screenshot → "Capture mode".

Files — Attached screenshots

Up to MAX_FILES per request
Shows the screenshots you have attached but not yet sent. Click to expand the popover — from there you can preview, remove individual files, or clear them all.
  • Files are cleared automatically after the AI request completes.
  • Drag-and-drop into the AissenceAI window also works on macOS and Windows for image files.

Web Dashboard

Opens https://www.aissence.ai/dashboard/personalize in your default browser
Jumps you straight to the Personalize tab of the web dashboard, where prompt presets, response settings, and your résumé context live. The desktop app reads from the same account, so changes you make here flow into your next session.
  • This is also the right place to update your billing plan (Settings → Billing) or your stored API keys.
  • Closing the browser tab does NOT close the desktop app.

Gear — App settings

Opens the in-app Settings (a.k.a. Dev-Space)
Opens the full Settings window (formerly called Dev-Space). The app resizes to a comfortable 1100×820 layout and re-centers on screen so every panel is visible without dragging.
  • If you previously saw a tiny / cropped version of Settings, that bug is now fixed — the gear always opens a full-size, centered window.
  • Inside Settings you will find: AI Providers, Speech Provider, Audio Devices, VAD config, Screenshot config, Shortcuts, Privacy and Appearance.

Close (X)

Quits the desktop app cleanly
Stops all active captures, releases the mic / system-audio devices, and closes the app. If Privacy Mode is OFF, transcripts and AI responses are wiped at this moment.
  • On macOS, ⌘Q in the menu bar does the same thing.
  • You can also use the global Quit shortcut from Settings → Shortcuts → Quit AissenceAI.
Reference

Settings panel reference

Click the gear icon in the toolbar to open Settings. The window expands to a full 1100×820 view (no manual resize needed). Here is every panel and what each option does.

App settings

Global behaviour: mini mode, click-through, opacity, always-on-top, app-icon visibility.

  • Always on top
    Keeps the AissenceAI window above every other window so it never gets covered by a Zoom or browser pop-up.
  • Click-through
    Makes the window ignore mouse clicks (they pass through to the app behind). Useful for stealth note-taking.
  • Mini mode
    Collapses the toolbar to its narrowest possible form — only the active capture state is visible.
  • Window opacity
    Slider from 30% to 100% so you can fade the toolbar against your wallpaper.
  • App-icon visibility
    Hides AissenceAI from the dock (macOS) / taskbar (Windows). It still runs and global hotkeys still work.

LLM / AI Providers

Choose which AI model answers your questions. Pick AissenceAI's hosted model (your subscription) OR plug in your own provider key.

  • Use AissenceAI backend
    The default. Routes through your subscription. No API key needed.
  • OpenAI / Anthropic / Gemini key (BYOK)
    Paste your own API key in Settings → AI Providers. Requests then bill against your key.
  • Model picker
    When BYOK is on, a dropdown lists every model your key can access. Pick the one you want for both chat and screenshot analysis.
  • System prompt
    A persistent instruction prepended to every AI request. Edit it from the web dashboard (Personalize → System Prompt) and the desktop app picks it up automatically.

Response settings (right panel)

Controls how the AI answers — applied to every response without retyping the instruction each time.

  • Length: Short
    Forces 2–4 sentences. Best when you only need a quick fact or a one-line answer mid-interview.
  • Length: Medium
    Forces 1–2 paragraphs (4–8 sentences). The everyday default for behavioural questions.
  • Length: Auto (default)
    AI picks the right length based on question complexity. Keep this on unless you have a specific need.
  • Language
    Forces the AI to reply in your chosen language: English, Spanish, French, German, Italian, Portuguese, Dutch, and more. Defaults to English.
  • Conversation Mode (⌘/Ctrl + K)
    Toggles between single-shot answers and a back-and-forth chat thread that remembers context across messages.

Screenshot configuration

What happens when you press the screenshot button or its global hotkey.

  • Capture mode: Screenshot
    Captures the entire visible window region behind AissenceAI. Toolbar shows the laptop icon.
  • Capture mode: Selection
    Lets you draw a rectangle around exactly what you want. Toolbar shows the cursor / mouse-pointer icon.
  • Processing mode: Auto
    After the screenshot is taken, AissenceAI immediately sends it to the AI with your auto-prompt — answer streams back in seconds.
  • Processing mode: Manual
    Screenshot is attached to the input box. You type the question and press Enter.
  • Auto prompt
    The text the AI receives in Auto mode. Default: "Analyze this screenshot and provide insights". Edit it for coding interviews ("Explain this code and suggest improvements"), system design ("What pattern is shown here?"), etc.
  • Enabled toggle
    Master on/off switch — when off, the screenshot button shows the Selection icon and Auto mode is bypassed.

Audio — System audio

How AissenceAI hears the interviewer. Configure your VAD or Continuous capture, the silence threshold, and which output device to monitor.

  • Capture mode: VAD
    Voice Activity Detection. Audio is only sent to the transcriber when speech is detected — no API calls during silence. Lowest cost, lowest latency.
  • Capture mode: Continuous
    Stream every millisecond. Use for noisy environments, soft-spoken interviewers, or interviews with lots of background music.
  • System Audio Response Speed
    Slider 5–60 seconds (default 20s). Controls the maximum length of each audio chunk before it is automatically sent for transcription. Shorter = faster response but may cut off long sentences. Longer = more context per clip.
  • VAD sensitivity
    Slider 0.0–1.0. Lower = picks up quiet speech but more false positives. Higher = ignores quiet speech but cleaner segments.
  • Silence threshold
    How long of a pause counts as "end of utterance" before the AI is asked to respond. Default ~800ms.
  • Output device monitoring
    Which audio device AissenceAI listens to. By default, your OS default playback device.

Audio — Microphone

Your voice input device, transcription quality, and the interplay with VAD.

  • Input device
    Pick the exact mic in the dropdown. Do not leave on "Default" if you have multiple mics — Windows can switch them randomly.
  • Microphone sample rate
    Read-only display of the negotiated sample rate (typically 16000 Hz). Useful when reporting bugs.
  • Push-to-talk vs always-on
    In VAD mode the mic auto-segments. In Continuous mode it streams until you toggle off.
  • Microphone test
    Live waveform visualisation so you can confirm audio is actually arriving before joining a call.

Shortcuts

Every global hotkey is rebindable. Defaults are listed in the Full keyboard shortcut reference below.

  • Edit a binding
    Click the keycap next to an action, press your new combo. Conflicts are detected automatically and shown in red.
  • Disable a binding
    Toggle the switch on the right of any row to turn that hotkey off without losing the binding.
  • Reset to defaults
    One-click button at the bottom of the panel restores every shortcut to the recommended out-of-box bindings.

Privacy

Who keeps a copy of your conversations and for how long.

  • Store conversation history
    OFF by default. When OFF, transcripts and AI responses live only in this app session and are wiped when you close the window. When ON, every chat is saved to your web dashboard.
  • Anonymous usage telemetry
    Off by default. Toggle on to share aggregate, non-identifying usage so we can prioritise features.
  • Auto-delete screenshots after 24h
    Recommended for shared machines. Anything captured locally is purged after a day even if you forget.

Appearance & Stealth

Window look and screen-share invisibility.

  • Stealth mode (Cmd/Ctrl + I or Cmd/Ctrl + Shift + H)
    Hides the AissenceAI window from screen-capture APIs that Zoom, Teams, Meet, Webex, and OBS use.
  • Theme
    Light / Dark / System. Defaults to System.
  • Cursor
    Set to "invisible" to hide the cursor only when over the AissenceAI window — clicks still register.
  • Font size
    Use the – and + buttons to adjust AI response text size (12–24px, default 14px). Saved automatically and persists across sessions.

Visit Web Dashboard

Opens https://www.aissence.ai/dashboard/personalize in your default browser.

  • Personalize
    Edit your system prompt, response presets and résumé context. The desktop app picks them up on the next request.
  • Activity
    Replay a saved transcript. Only available with Privacy Mode ON.
  • Billing & subscription
    Upgrade, downgrade, change payment method.
  • DownloadSimple new versions
    Always grab the latest installer from /dashboard/desktop-app.

Quit AissenceAI

Stops all captures, releases devices, removes the global hotkey listeners.

  • Quit (in sidebar footer)
    Same as the close (X) button on the toolbar. Clean shutdown.
  • Force quit (OS)
    Cmd+Option+Esc on macOS or End Task on Windows — should never be needed in normal use.
Reference

Full keyboard shortcut reference

Defaults straight from the source. Every binding can be remapped in Settings → Shortcuts.

ActionmacOSWindows / LinuxWhat it does
Toggle Stealth Mode⌘+ICtrl+IHides / shows the overlay without closing the app or stopping captures.
Toggle Window⌘+\Ctrl+\Quickly hide/show the main window. Great when an interviewer asks you to share screen unexpectedly.
System Audio⌘+⇧+MCtrl+Shift+MStart / stop system-audio capture (the headphones icon).
Voice Input⌘+⇧+ACtrl+Shift+AStart / stop your microphone capture.
Screenshot⌘+⇧+SCtrl+Shift+STake a screenshot in whichever mode (full / selection, auto / manual) is configured in Settings → Screenshot.
Refocus Input⌘+⇧+ICtrl+Shift+IBrings AissenceAI forward and places the cursor in the input box, even if another app is focused.
Toggle Dashboard⌘+⇧+DCtrl+Shift+DOpen / close the in-app Dashboard window (chats, history).
Move Window⌘+↑↓←→Ctrl+↑↓←→Hold the modifier and tap an arrow to nudge the overlay. Hold to move continuously.
Conversation Mode⌘+KCtrl+KInside the response panel: toggle between single-shot answers and a continuous chat thread.
Dismiss / CancelEscEscCloses the AI suggestion popover or cancels an in-progress selection.
Conflicts with another app? Open Settings → Shortcuts, click the keycap, and press your new combo. AissenceAI shows a red warning if the new combo collides with another action.

Frequently asked questions

Does the interviewer see anything from the desktop app?

No. With Stealth Mode on, the window is excluded from screen-capture APIs that Zoom, Teams, Meet, Webex, and OBS use. The interviewer sees only your webcam and your shared screen content.

Is my interview audio sent to the cloud?

Audio is sent to a transcription provider in real time so the AI can respond. With Privacy Mode OFF (the default), nothing is stored after the session ends. With Privacy Mode ON, the transcript is saved to your dashboard.

Can I use my own OpenAI / Anthropic / Gemini API key?

Yes. Open Settings → AI Providers and paste your key. The app will route requests through your key for that provider. Your subscription still controls feature access.

Does the app work offline?

No — transcription and AI responses both require an internet connection. Use a wired connection or strong Wi-Fi for the lowest latency.

I just upgraded my plan. Why does the app still say "Subscription required"?

Click the "I've upgraded" button in the upgrade banner — it forces a fresh subscription check. If that does not work, fully close the app and reopen it; the subscription state will refresh on launch.

Ready when you are

Open your dashboard to grab the installer, or read the dedicated stealth-mode guide before your next live interview.

Get the installerRead the Stealth Mode guideSee the changelog