Hook Generator
Get openings in several different shapes — each with the first visual, and the promise the video then has to keep.
Describe the video and press Write the Hooks.
Four to six openings, each a different structure.
Writing openings…
This usually takes ten to twenty seconds.
Got your opening? Draft the rest.
Write the script →A hook is a line and a shot, not just a line
In a faceless video the opening is two things happening at the same time: what the narrator says, and what is on screen while they say it. Most hook tools return only the line, which leaves the harder half to you — and a strong line over a generic shot of a laptop is a weak hook. Every opening here comes with a first visual described concretely enough to go and find or generate: a subject doing something, framed a particular way.
Why each hook states what it promises
A hook takes on a debt. It tells the viewer what they are about to get, and the video has to pay that back. If it does not, they leave at second forty rather than second five — which hurts more, not less, because the view has already been counted and the drop is now measured against a video that looked like it was working.
So each hook names its promise. Read that line before you read the hook itself. If the only way to pay it off is with information you do not have, that is the signal to pick a different opening rather than to rewrite the video around it.
The "most interesting thing" field is the one that matters
It is optional, and it changes the output more than every other field combined. Without it the model builds openings on the shape of your topic, because inventing a surprising detail would mean inventing a fact — which it is explicitly forbidden to do. With it, the hooks are built on the thing that actually makes your video worth watching.
If you cannot fill it in yet, that is worth noticing on its own. A video with no surprising specific in it is a video with no hook, and no tool can supply one honestly.
Long-form and Shorts are not the same job
A long-form opening has ten to fifteen seconds and can afford a sentence of setup. A Short has about two, and the visual has to land at the same moment as the first three words. Switching the format here does not just shorten the line — it changes which structures work at all, which is why the two settings return genuinely different sets.
Where it sits in the workflow
Write the hook before the script, not after. It is the one part of a video that decides whether the rest gets watched, and having it first tells you what the script has to deliver. From here, the Script Generator drafts the narration, the Scene Splitter turns that into a shot list, and the Title Analyzer checks that the promise survives the browse page.
Common questions
Why are the hooks so different from each other?
Because asking a model for five hooks gets five rewordings of one idea. Each hook here is written to a different structure — a cold open, a contradiction, the stakes stated plainly, a concrete image before any explanation — so you are choosing between genuinely different openings rather than between phrasings. Pick by shape first, then edit the wording into your own voice.
Will it invent a statistic to make the hook stronger?
No. The model is told never to state statistics, dates, amounts, records or quotations it cannot stand behind, and the result is rejected outright if a percentage or a currency figure appears in a spoken line. A hook is the single most dangerous place for an invented number, because it is the line most likely to be recorded word for word and checked last.
Why does every hook come with a shot?
Because in a faceless video the opening is two things at once — what the narrator says and what is on screen while they say it. A strong line over a generic laptop shot is a weak hook. The first visual is described concretely enough to search a stock library for or hand to a generator, rather than as "something eye-catching".
What is the "what it promises" line for?
A hook takes on a debt. It tells the viewer what they are about to get, and the video has to pay it back or they leave at second forty instead of second five — which costs more, because the view has already been counted. If a hook can only be paid off by information you do not have, that is the signal to pick a different one rather than to write the video around it.
Does it use one of my daily AI generations?
Yes. The free allowance is shared across every AI tool on the site rather than given separately to each one, because each generation costs real money whichever tool spends it. Everything that runs in your browser stays unlimited.