Tap command twice, and it sees your window.
An Appshot is the fastest way to show the assistant what you are looking at. Double-tap the command key and the window in front of you is captured and staged in the chat, named after the app it came from. Then ask.
abstract
Most of what you want from an AI at work is about something already on your screen: the Slack reply you are about to send, a client email in a language you half know, an error that means nothing to you. Describing it is slower than showing it. An Appshot captures the frontmost window in one gesture and hands it to the chat, where you ask in plain words: tighten this, translate this, explain this. The image is processed on your Mac and never leaves it.
keywords: double-tap ⌘ · the window, not the screen · ask about what you see
theorem 5.1 (the capture)
One gesture, no framing.
Proof. No crosshair to drag, no region to pick. Double-tap command and the frontmost window is the shot, whatever window that is, one of Tesseract's own included. It lands in the composer labeled with the app and window title, ready for a question.
in plain terms
- ·double-tap ⌘
- ·the frontmost window, whole
- ·staged in the chat, named
theorem 5.2 (the workday)
Built for the message you are about to send.
Proof. The everyday use is words. A reply you want a second opinion on before it goes to the whole channel. A thread in a language you read slowly. A dialog full of jargon. Appshot it, ask, and the answer arrives in the chat beside your work. Nothing is typed into the other app; you stay the editor.
in plain terms
- ·proofread before sending
- ·translate what you see
- ·explain what is on screen
theorem 5.3 (the patience)
The shot waits for you.
Proof. A staged Appshot is part of your draft, and the draft survives you: switch conversations, start a new chat, come back later, and it is still there until you send or discard it. And it is an image and nothing more. No text is scraped from the window behind your back; the assistant sees exactly what you saw.
in plain terms
- ·survives switching chats
- ·an image, nothing scraped
- ·sent only when you send
fig. 01: from window to question. The shot is staged, named, and waiting for you.
the gesture
tap ⌘ command twice · changeable in Settings
what is captured
the frontmost window, whole, as an image and nothing more
the permission
Screen Recording, granted once, applied after a relaunch
where it goes
the chat composer on your Mac · never anywhere else
§ endsee for yourself
Appshot is shipped and in daily use. It needs a vision-capable model from the chat paper and the one permission macOS requires, and the shot never leaves your Mac.
next in the series: paper no. 06 · the server →