How to pause a demo on the click that matters

Last updated 2026-09-15 · 8 min read

What is the short version?

Whoever recorded the demo knew where every button was. The viewer does not, and a screen that changes the instant it is clicked gives them nothing to look at — the information was on screen for a third of a second.

1. Go to Vizard Agent with the recording.
2. Name the moments the viewer has to read.
3. Say how many seconds each one needs.

What do you need before you start?

The recording and a list of the moments that carry information. Vizard Agent holds what you name, and the useful list is not "the important bits" — it is the specific clicks, screens and figures a viewer is expected to take in rather than watch go past.

• The recording. Uncut, ideally.
• The moments to hold. By what is on screen.
• How long each needs. Three seconds is a good default.
• Whether somebody is speaking over it. It changes the timing.
• Which passages are scrolls. Those get slowed, not held.

What do you type into Vizard Agent?

Name what is on the screen to Vizard Agent rather than giving it a timecode. Timecodes shift every time the edit changes, whereas "the moment the cursor clicks Visual Analysis" survives a re-cut and means exactly the same thing to both of you afterwards.

Prompt

Variants worth knowing:

• "Where the cursor clicks X." Survives a re-cut.
• "So the speaker can explain it." Sets the length.
• "Slow it rather than extend the cut." Two different fixes.

What does Vizard Agent actually do?

Here is the sequence Vizard Agent worked through on a real product walkthrough built from screen captures. Notice that a hold and a slow-down get treated as two different tools for two different problems rather than as one pacing setting applied everywhere.

1. Reads the action logs of each clip to find what happens where.
2. Checks the screen where the hold is meant to go.
3. Adds the hold at the frame the click lands on.
4. Rebuilds the video with that timing.
5. Verifies the hold in the finished cut.
6. Reviews the held frame to confirm it is the right one.
7. Applies motion-smoothed slow motion to the scrolling passages.
8. Checks the slowed scroll for smoothness and artefacts.
9. Extends the clip's duration where slowing alone is not enough.

Step three is the one that needs precision. A hold placed a few frames before the click freezes the screen the viewer is leaving rather than the one they are arriving at, so Vizard Agent puts it on the frame where the new screen is fully drawn.

Step seven is why scrolls are a separate case. A scroll carries its meaning in the movement, so freezing it destroys the thing you wanted shown — slowing it keeps the motion and gives the eye time to travel with it.

Step eight is not optional with slow motion. Interpolated frames can smear text, which is the one thing a demo cannot afford, so Vizard Agent looks at the slowed passage rather than trusting that slower is automatically more readable.

What does the result look like?

A demo that stops where the information is. It runs longer than the raw recording, because the recording was made at working speed and the video is made at reading speed, and that difference is the whole point.

Vizard Agent tells you how long each hold runs, so you can trade seconds if the runtime is fixed.

When does this not work well?

Holds cost time and attention. Too many and the demo stalls, which is its own failure — and a screen that is genuinely dense may need a graphic or a callout rather than three more seconds of the same picture.

• Too many holds. The demo stops moving.
• Very dense screens. Three seconds will not be enough.
• Fast social formats. There is no room for a hold.
• Interpolated slow motion on text. It can smear.
• Live narration already recorded. The timing is fixed to it.

How do you fix a result that came back wrong?

Say whether it is too short or the wrong frame. Because Vizard Agent places holds at a named screen, "it holds on the previous page" and "three seconds is not enough" are different corrections and naming the right one avoids redoing both.

• "It holds on the wrong screen." The frame moves to the click.
• "Still too fast to read." A longer hold, or a callout.
• "The scroll looks smeared." Less interpolation, or a cut.
• "It drags now." Fewer holds, not shorter ones.

How does Vizard Agent compare to doing it yourself?

By hand a demo is reviewed by the person who recorded it, who can read every screen because they already know what it says. The pacing feels fine to them and unreadable to everybody else, and the note that comes back is "it made me dizzy".

Common questions

How long should a hold be? About three seconds. Vizard Agent will lengthen it if you are speaking over it.

Can I name the moment without a timecode? Yes, and Vizard Agent prefers it — timecodes move when the edit changes.

What about scrolling passages? Vizard Agent slows those rather than holding them; the movement is the content.

Will slow motion blur the text? It can. Vizard Agent checks the slowed passage for exactly that.

Does the video get longer? Yes. Reading speed is slower than working speed.

Can it add a callout instead? Yes, when the screen is too dense for a hold to help.

How many holds are too many? Enough that the demo stops moving. Vizard Agent will say when it does.

What if there is a voiceover? Then Vizard Agent matches the hold to what is being said over it.

Can it hold and zoom at once? Yes, though a zoom on a held frame reads as emphasis rather than as a pause.

Will the cursor keep moving? Not during a hold. Vizard Agent freezes it, which is what makes the pause read as deliberate.

Can it find the moments itself? It can propose them from the action log; the choice stays yours.

Does it work on phone recordings? Yes, and Vizard Agent finds those usually move faster still.

What about the audio during a hold? Vizard Agent lets it continue, so the speaker keeps talking over the still screen.

Why did it feel fine when I watched it? Because you knew what every screen said before it appeared.