How to review your candidate moments as one playable strip

Last updated 2026-09-10 · 8 min read

What is the short version?

A shortlist of forty moments from a back catalogue is a list of timecodes. Reading it tells you nothing about which ones land, because what makes a line work is the pause before it and the way it is said.

1. Go to Vizard Agent with the shortlist.
2. Ask for one strip of all the candidates, with sound.
3. Watch it once, then say which survive.

What do you need before you start?

The candidates and somewhere they can be gathered from. Vizard Agent needs enough of each moment to judge it — a line cut to its exact words is unjudgeable, because the delivery lives in the second before and the second after.

• The shortlist. Timecodes or a search result.
• Access to the sources. Files, or the catalogue.
• A little handle on each side. Two seconds is usually enough.
• Roughly how many you want in the end. It shapes the cut.
• Whether you will watch it or someone else will. Say so.

What do you type into Vizard Agent?

Ask Vizard Agent for the strip explicitly. Left to its own devices a review pass tends to arrive as a sheet of still frames, which is the right tool for judging framing and composition and quite the wrong one for judging whether a line is actually funny.

Prompt

Variants worth knowing:

• "With sound." The whole point.
• "In order, with a little handle." So each one is judgeable.
• "And the transcript." So you can name them afterwards.

What does Vizard Agent actually do?

Here is the sequence on a real compilation cut from an entire podcast back catalogue. The strip goes together early, long before any styling decisions, and it is what turns a list of forty candidates into a running order of nine.

1. Builds the catalogue of what exists.
2. Searches the transcripts for candidate moments.
3. Reads the exact wording and timings of the shortlist.
4. Downloads each candidate as a segment.
5. Checks the segment lengths are what it asked for.
6. Assembles them into one review strip.
7. Uploads the strip so it can be watched properly.
8. Transcribes and watches it end to end.
9. Reads the strip's own transcript back.
10. Cuts the shortlist down to the ones that hold.
11. Only then lays out the clip list and timings.

Step six is the cheap step that saves the expensive one. Assembling a strip costs a few minutes; discovering on the finished, captioned, music-mixed edit that four of the moments are flat costs a rebuild.

Step eight matters because watching is not the same as checking. Vizard Agent watches the strip as a viewer would, in order, with the sound, rather than sampling frames out of it — which is the only way to tell that a line needs the sentence before it.

Step eleven is the discipline. Nothing gets styled, timed or scored until the shortlist has survived a watch-through, so no effort is spent dressing a moment that is going to be dropped.

What does the result look like?

One file, a few minutes long, containing every candidate moment in order with its sound intact. You watch it once and know which ones are in and which are not, and so does anybody else you send the link to.

Vizard Agent keeps the strip's transcript beside it, so an instruction like "drop the third one and the one about the invoice" is completely unambiguous.

When does this not work well?

A strip is for judging performance. It is the wrong tool for judging picture quality, continuity or framing, and it can mislead about pace because forty moments back to back feel faster than they will in a finished cut.

• Judging framing. Still frames are better for that.
• Very long candidates. The strip stops being watchable.
• Silent footage. Then a contact sheet is the right tool.
• Pace judgements. Everything feels fast in a strip.
• Hundreds of candidates. Narrow by text first.

How do you fix a result that came back wrong?

Refer to the moments by their words rather than their positions. Because Vizard Agent transcribed the strip itself, naming a line is exact in a way that "the fourth one, or maybe the fifth, the one after the laugh" never quite manages to be.

• "Cut the one about the invoice." Named, not counted.
• "That one needs more before it." The handle gets longer.
• "Add three more from episode 12." A new strip.
• "They all feel the same." Vizard Agent varies the selection.

How does Vizard Agent compare to doing it yourself?

By hand the shortlist gets reviewed by scrubbing between forty timecodes across a dozen files, which takes an afternoon and gives you no sense of how the moments sit next to each other. Most people skip it and find out in the edit.

Common questions

Why not use a contact sheet? Because stills cannot tell you whether a line is delivered well. Vizard Agent uses both, for different jobs.

How long should the strip be? Short enough to watch in one go. Vizard Agent will say if it is not.

Can I share it? Yes. Vizard Agent uploads it, so it is one link.

Does it include captions? It can, but the transcript Vizard Agent supplies alongside is usually enough.

Will it cut the shortlist itself? Vizard Agent will recommend. The choice stays yours.

Can it group by theme? Yes. Tell Vizard Agent the grouping you want in the strip.

What about clips with no speech? Then frames are the better review. Vizard Agent will say so.

How much handle does each moment need? Enough to judge the delivery, so Vizard Agent leaves a second or two either side.

Can it do this across a whole catalogue? Yes, and that is where Vizard Agent doing it pays off most.

Does the strip get thrown away? It can be kept. It is useful for the next compilation.

Will it re-download everything? No. Vizard Agent cuts the strip from what it already pulled.

Can somebody else pick? Yes. That is the reason it is one shareable file.

Does the order in the strip matter? Not yet. Vizard Agent decides running order after the shortlist.

Why does everything feel fast in it? Because there is no room between the moments in a strip. The finished cut will breathe, so do not judge pace from it.