Does Jarvis Store My Screenshots on Its Servers?

No. Jarvis (getjarvis.eu) does not keep a library of your screenshots. When you invoke screen awareness with Cmd+/ on macOS or Ctrl+/ on Windows, Jarvis captures a single snapshot of the active window, sends it over TLS to the EU backend, passes it to the model that fits the task (frontier models from Anthropic, OpenAI, or Google) to read for that one answer, and does not retain it as a permanent file. Screenshots are never used to train any model. There is no continuous screen recording and no rolling capture buffer. Jarvis (the floating-bar desktop AI for macOS 12+, Windows 10+, and Linux) treats each capture as transient request context. The persistent memory layer stores facts you choose to keep, not raw images. Pricing is free to start, then $16/month for Pro.

A screenshot only happens when you explicitly ask Jarvis to look at your screen. There is no background daemon snapping your display on a timer, and no hidden recording while the bar is idle. If you type a question that needs no visual context, Jarvis takes no image at all. This is the opposite of meeting-recorder or always-listening tools: capture is a deliberate, per-request action you trigger with the hotkey, so you always know when your screen is being read.

Once captured, the image is encrypted in transit and processed to produce your answer, then dropped from the working request. It is not filed into an album, not indexed for search, and not fed back into frontier models from Anthropic, OpenAI, or Google for training. What can persist is a fact you ask Jarvis to remember, for example a number it read off a dashboard, stored as text in the user-controlled memory you can inspect, edit, or clear at any time, never as the underlying picture.

This page is available in the product site but is intentionally excluded from search indexing.

Privacy & data