Output action

Path: Settings → Text post-process → Output action.

Semantic destination is DictationOutputAction (orthogonal to how text is injected):

Options

Action	Behavior
Insert at caret	Default: inject into the focused field; if none, show a Copy card
Insert & send	Insert, then simulate Return (chat / agents)
Copy to clipboard only	No insert — clipboard only
History only	Local history only — no insert or copy

Also (only when inserting):

• Keep transcript on clipboard after insert
• Restore previous clipboard after paste (priority follows in-app copy)
• Insertion method: smart / Accessibility / paste / type (default: smart)

Quick Send vs Insert & send

	Quick Send	Insert & send
Where	Settings → General	Output action
When	Return while recording	Every finished dictation
Use	Occasional “speak and send”	Always behave like chat IME

See Hotkeys & recording.