How it works

Drag a file into the app, choose one from your library, or take a photo with your camera. Uploads are validated before anything else happens.

PNG, JPG or WEBP — up to 10 MB

Uploading just stores the image securely. When you press Analyze, our server sends it to the AI vision model and runs three phases:

• Reading screenshotThe model reads the text and layout visible in the image.
• Extracting actionsIt pulls out actions, deadlines, people and priorities — nothing else is added.
• FinalizingResults are checked, saved to your history and shown to you.

Vision models can guess. ActionShot is built not to, so the extraction follows four rules:

• ✓Your original wording and language are preserved — tasks are not rewritten.
• ✓Nothing is invented. If a screenshot has no clear request, you get an empty result instead of filler tasks.
• ✓Relative dates such as “tomorrow” or “by Friday” become real dates only when they can be resolved safely; otherwise the deadline stays exactly as written.
• ✓Low-confidence extractions are flagged as uncertain so you can check them against the original screenshot.

Open the result and work through it: edit any action, mark it complete, copy it, or delete what does not apply. When the list is right, export it:

ExportsPlain textChecklist (Markdown)CSV (spreadsheet)Calendar (.ics)