Smart Extract is the fastest way to capture *content* from your iPhone screen into a working document.
How it works
- Mirror your iPhone or open a recorded session.
- Hold ⌘+Alt on a Mac, Ctrl+Alt on Windows.
- Drag a rectangle around whatever you want to capture.
- Release.
The selected region is sent to Claude, which returns formatted Markdown. The Markdown is automatically placed on your clipboard *and* in a result card in the sidebar.
What you get
Smart Extract preserves structure, not just text. If your iPhone screen has:
- A table → comes back as a real Markdown table
- A bulleted list → comes back with
-bullets and proper nesting - Code → wrapped in triple-backticks, indentation preserved
- A receipt or itinerary → labelled key-value pairs
- A chart → the data behind it, as a short table
This is genuinely different from OCR. Plain OCR gives you the *text* but loses the structure. Smart Extract reconstructs it.
When to use Smart Extract vs Copy Text from Screen
Use Smart Extract when the content has *structure you'd lose with plain text* — tables, lists, code blocks.
Use Copy Text from Screen (right-click, or ⌘⇧-drag on Mac / Ctrl+Shift-drag on Windows) when you just need the raw text — a single-line error message, a name, an address. It's free and never touches your tokens. On a Mac it runs on-device via Apple Vision; on Windows it's built-in text capture.
What this costs
One AI token per drag. Drag a tiny region, drag the whole screen — same cost.
Tips
- The Markdown lands on the clipboard ready to paste into Notion / Obsidian / VSCode / a Jira ticket / wherever.
- If the result has an obvious mistake, hit Re-extract (↻) in the result card to spend another token retrying. The second response tends to be better when the first was off.
- Smart Extract works on recorded sessions too — drag from any snapshot.