Cue

Say the thing, and your Mac does it.

A voice-first assistant that reads what you're working on and acts on it. Press Option-Space or just say its name, there's no window to find and nothing to switch to.

Coming to the Mac App Store

See how it works
Cue's assistant panel open over Mail on macOS

A colleague who was already watching.

Most assistants make you stop, switch context and phrase your thoughts for an empty chat window, then copy the answer back. Cue already knows the app you're in, the text you highlighted and the thing you just copied, so you talk, it acts, and the work keeps going.

Say its name, or press a key

Say "Cue" or hit Option-Space and just talk. There's no app to switch to and no window to find, the orb floats over whatever you were doing and gets out of the way when you're done.

Sees what you're working on

Cue knows the app in front of you, your selection and your clipboard, and reads the screen when you ask about it, so you never have to explain where you are before asking for something.

Actually does the thing

Claude decides and Cue acts, opening apps, putting windows where you want them, searching the web and reading pages for you. Asking for something should get it done.

You can talk over it

Start speaking while Cue is mid-answer and it stops to listen, the way a person would. A cough or an "mm" doesn't count, real words do, and it remembers the part of the reply you actually heard.

Native Mac, top to bottom

Swift and SwiftUI, a Metal orb, and a panel that never steals focus from whatever you were typing into. It feels like part of macOS because it was built the way Mac apps are built.

Your keys, nobody in between

Speech is recognized on your Mac whenever it supports that, and only the text goes to Claude, on your own Anthropic key. There's no Cue server in the middle, so nothing about you lands with us.

How Cue thinks

Each stage is cheap until the one before it earns the next, so nothing wakes Claude up for nothing. Your battery, your API bill and your attention all get respected.

  1. 1. Passive sensors

    The app in front of you, your selection and your clipboard. Cheap, local and always on, with no screen capture, no mic and no model calls.

  2. 2. Screen reading

    ScreenCaptureKit plus Vision text recognition, only when you ask about the screen. The text is read on the Mac and the image never leaves it.

  3. 3. Trigger

    Option-Space, or the wake word if you switch it on. Later, a small on-device scorer gets to decide when something is worth bringing up by itself.

  4. 4. Transcription

    Apple's speech recognition turns your voice into text, on the Mac whenever it can, and the end of your turn is judged from the words, so a pause after "and" doesn't cut you off.

  5. 5. Reasoning

    Context, transcript and tools go to Claude, and Cue starts speaking from the first sentence of the reply while the rest is still streaming in.

  6. 6. Action

    Claude picks a tool and Cue runs it, then tells you what happened. Read-only lookups run side by side, so a question that needs three of them doesn't keep you waiting.

Deliberately narrow

The scope is deliberate, and keeping it tight is how it stays good.

Voice first

Chat is there as a fallback, the voice loop is the main way in and everything is designed around it.

Lives in the OS

Cursor has code editors covered and does it well, Cue sits a level up and works across every app on the Mac.

Made for working hours

Granola has meetings covered, Cue is for the hours you spend actually doing the work.

Listens when invited

It hears you when you press the hotkey, or when you say its name if you turn the wake word on, and your voice is never recorded or stored.

Mac only

Mac first and Mac only, at least until the shape of the product is right.

Sold on the App Store

Sandboxed and sold through the Mac App Store, with a free 14-day trial and a one-time unlock after it.

Boring where it counts

Platform-native, mostly Apple frameworks and one LLM vendor. Small surface area, long shelf life.

Swift + SwiftUI, macOS 14+, Speech framework, Anthropic Claude API, AVAudioEngine, ScreenCaptureKit, Vision (text recognition), Metal, Carbon RegisterEventHotKey, NSPanel (non-activating), Accessibility API, StoreKit 2

The Mac assistant that actually does stuff.

Free for 14 days on the Mac App Store, then a one-time unlock. It runs on your own Anthropic API key, so Anthropic bills what you use directly and there's no subscription.

macOS 14 Sonoma or later.

Coming to the Mac App Store

Ask a question