The hook widget

• Hook lines: enter multiple angles (one per line), or generate them from the briefing. Variants rotate through them, so ten renders don't all open with the same sentence.
• Three modes:
  • Overlay — the line sits on the video for its time window, then fades. The video's length is unchanged.
  • Card — a dedicated card segment (color or image/video card) is placed in front of the video: the hook text on a clean surface, then your video starts.
  • Blur — the opening seconds of your own video, blurred (frozen or moving), with the hook text on top — then it un-blurs into the real start.

Click to enlarge×

Fix the offer before adsThe offer does the sellingYour ads aren't the problem

Styling and voiceover

• Styling: color, size, font, shade — plus an optional motion title: the line animates in instead of standing still. Twelve looks, chosen from a tile grid that plays each animation on hover. The brand's accent color is the default; a stack can override accent and font.
• Voiceover : in card and blur mode, an AI voice can speak the hook line over the otherwise silent pre-roll (bring your own TTS key; per-brand voices; volume slider). Spoken lines are cached — twelve variants sharing a hook line cost one TTS call — and the editor plays the voice during preview playback. A voiceover is synthetic speech, so the post usually needs the platform's own AI label — how to set it.

Where a hook line comes from

Three sources, in this order:

1. The per-language pool for the brand's content language, when the stack has one.
2. The flat list on the Hook tab.
3. The line stored on the hook widget for that brand — what an assistant writes over MCP when each brand needs its own wording.

The Hook tab says so when it applies: with per-brand lines in play it shows "This stack's hook lines are stored per brand on the hook widget", and a line typed into the list replaces them for every brand. When pools exist for some languages but not for a brand's, the tab names the gap — those brands fall back to the flat list.