Vizard Agent

How to cut a testimonial that shows a change

Last updated 2026-08-30 · 7 min read

Upload the interview and tell Vizard Agent the clip must show a concrete change rather than a recommendation. Vizard Agent transcribes everything word by word, finds the sentences that describe a before and an after, checks the framing at each one, and cuts a short testimonial around that arc.

What is the short version?

Most testimonials are somebody saying the course was great. The ones that work say what they could not do before and can do now, and those sentences are usually buried in the middle of an interview rather than in the answer to "would you recommend it".

  1. Go to Vizard Agent and upload the interview footage.
  2. Say the clip must show a concrete change, not a recommendation.
  3. Name what it is promoting.

What do you need before you start?

The unedited interview and the offer it supports. Vizard Agent reads the whole transcript to find the change, so no timestamps are needed, but you should know what you are selling because that decides which change is the relevant one.

What do you type into Vizard Agent?

Ask for the change rather than for the praise. Vizard Agent selects entirely differently when told to look for a before and an after, and that instruction is what stops the clip becoming a montage of somebody saying nice things.

Prompt

Cut these interviews into a [40-50] second vertical testimonial for [platform]. The goal is to create interest in [the offer] by showing a concrete change — not a simple recommendation. Readable captions, discreet instrumental music.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real testimonial cut. The step that most people skip is the sixteenth: it checked the faces and the backgrounds at each selected moment before locking the selection in.

  1. Checks the video's duration and format and builds a visual overview.
  2. Loads the testimonial playbook and checks the transcription options.
  3. Looks at an overview of the shots across the footage.
  4. Transcribes all the speech with word-by-word timings.
  5. Detects the shot changes and who is in each one.
  6. Reads the transcript to select the sentences that carry the change.
  7. Reads the visual overview alongside the text, and checks the frames at the chosen points.
  8. Checks the caption options and searches for very discreet instrumental music.
  9. Builds a grid of the selected frames, checks the faces and backgrounds in each, extracts the precise sentence timings, and picks the sharpest moment of the original end card.

Step 9's first half matters more than it sounds. A sentence can be exactly the right words and land on a frame where the speaker is mid-blink or a doorway is growing out of their head, and choosing on text alone means finding that out on playback.

What does the result look like?

The source interviews measure 1080x1920, HEVC, 30fps, 168.9 seconds in total. From that comes a forty-to-fifty second vertical testimonial: the sentences that describe what changed, captioned so it reads without sound, a discreet instrumental underneath, and the original end card reused at its sharpest frame.

The clip does not include anyone saying they would recommend it. That is deliberate — a specific before and after is more persuasive than a general endorsement, and it also survives scrutiny better.

When does this not work well?

A testimonial can only show a change the person actually described on camera, and the ethics of editing somebody's words into a marketing asset deserve real attention. These are the limits Vizard Agent works within and states plainly rather than glossing over.

How do you fix a result that came back wrong?

Say what the clip missed or misrepresented. Vizard Agent keeps the full transcript with its word timings, the shot detection, the selected frames and the framing checks, so a different passage can be cut without re-reading the whole interview.

How does Vizard Agent compare to doing it yourself?

By hand this means watching an interview twice, once for the words and once for the pictures, then discovering that the best sentence sits on the worst frame. Most people cut for the words and accept whatever the picture is doing.

By hand Vizard Agent
Finding the change Listen and hope Read from a word-by-word transcript
The framing Notice it later Faces and backgrounds checked per selection
Cut points Trim by ear Precise sentence timings extracted
The end card Freeze any frame The sharpest frame chosen

Common questions

What makes a testimonial work? A concrete before and after. Vizard Agent selects for that arc rather than for praise.

How long should it be? Forty to fifty seconds. Vizard Agent fits the arc in without letting it drift.

Do I need to mark the good bits? No. Vizard Agent reads the whole transcript and finds them for you.

Can it use several interviews? Yes. Vizard Agent selects across all of them and can build one clip or several.

Will it be readable without sound? Yes. Vizard Agent adds animated captions, which is how most people will watch it.

Why check the frame as well as the words? Because the sentence you want is often spoken while the person is looking away or mid-blink, and choosing on the transcript alone means discovering that only after the clip is built.

What about consent? Ask for this specific use. An interview given for one purpose does not cover another.

Can I keep the original end card? Yes. Vizard Agent finds its sharpest frame in the source footage and reuses it.

Should the person say the product name? Not necessarily. The change is what persuades, and Vizard Agent can put the name on the card.

Can I get several clips from one interview? Yes, and it is the efficient way to work. Vizard Agent already holds the transcript and the shot detection, so a second and third testimonial from the same sitting cost a fraction of the first.

Should the music be noticeable? No. Vizard Agent searches for something deliberately discreet, because a testimonial is persuasive through what the person says and a driving track makes the whole thing read as an advertisement.

Where should the offer be named? At the end, on the card. Vizard Agent keeps the spoken part about the person's own experience, which is more credible than a testimonial that turns into a sales line halfway through.

Does Vizard Agent check the finished clip? Yes. It reviews the selected frames and the caption timing before delivery.