How to swap the real data in a demo recording for clean demo names

Last updated 2026-09-15 · 8 min read

What is the short version?

A screen recording of a real session contains real names, and watching it back you will find perhaps three of them. The software knows about rather more than that: it remembered what you typed earlier and offers the whole list back in a dropdown two minutes later, for half a second, while you are clicking something else.

1. Give Vizard Agent the clean names you want shown.
2. Ask it to find every place a real one appears.
3. Ask for each replacement checked at full size.

What do you need before you start?

The recording and the names you want in its place. Vizard Agent replaces rather than hides, so the substitutes should look like something a real user would have typed — "Demo Dataset" reads as a demo, "XXXXXX" reads as a redaction.

• The recording. The full session.
• The replacement names. Plausible, not placeholders.
• What counts as sensitive. Names, emails, account numbers.
• Anything that may stay. Public product names.
• The finished look. So overlays match the interface.

What do you type into Vizard Agent?

Say "replace" and give Vizard Agent the new names to use. Asking for the names to be removed invites a blur instead, and a blurred field in the middle of a product demo tells the viewer there was something there worth hiding — which is the opposite of what a launch video is for.

Prompt

Variants worth knowing:

• "Rather than blurring them." The distinction that matters.
• "Including the autocomplete suggestions." The place people forget.
• "And the results table at the end." Names reappear late.

What does Vizard Agent actually do?

Here is the sequence on a real product walkthrough that had to be cleaned before launch. Nearly all of it is measurement: the replacing itself is trivial once you know exactly where each name sits and for how long.

1. Surveys the recording for every moment a name appears.
2. Views each name location before masking anything.
3. Inspects the exact fields at full resolution to cover them cleanly.
4. Builds a timeline of when each name is visible.
5. Measures the result rows where labels appear again.
6. Measures the modal geometry for each dialogue box.
7. Lays clean demo labels over every measured region.
8. Inspects each replacement and its transition at full resolution.
9. Masks the saved-name suggestion lists that appear on field focus.
10. Removes the remaining table exposure and re-checks the whole file.

Step nine is the one that catches people out. The autocomplete dropdown shows names you typed in an earlier take, so a recording can expose data that is not otherwise in it at all — and it only appears for the half second after a click.

Step six is why this is measurement rather than masking. A modal has its own geometry, and a label placed for the main window will sit a few pixels off inside the dialogue, which reads as a sticker rather than as the interface.

Step eight is not the same as watching it back. At playback size a replacement looks perfect; at full resolution you can see the original's descender poking out below the overlay.

What does the result look like?

A demo that looks like it was recorded with demo data. Nothing is blurred, nothing is missing, the labels sit inside the interface as though they were typed there, and Vizard Agent gives you the full-resolution frames of each replacement.

When does this not work well?

Replacing works on text that sits still. A name inside a scrolling list, a chart label that animates, or text at an angle is much harder to cover convincingly, and blurring or a re-record may be the better answer.

• Scrolling or animating text. The overlay has to move too.
• Names inside images. Not interface text, harder to match.
• Very small text. The replacement will not match weight.
• Names spoken aloud. Audio is a separate job.
• Data in a chart. Changing a label without the data misleads.

How do you fix a result that came back wrong?

Give the timecode and say what is showing. Vizard Agent has the measured regions, so a report like "the old name is visible under the new one at 1:12" points at one overlay rather than at the pass as a whole.

• "The old text shows underneath." That region is widened.
• "The label looks pasted on." It is re-measured against the modal.
• "A name appears in the dropdown." The suggestion list is masked.
• "You missed the download row." That table region is added.

How does Vizard Agent compare to doing it yourself?

By hand you re-record with clean data, which is the right answer and takes an afternoon of setting up an account that looks lived-in. The shortcut is blurring, which is quick and makes a launch video look like evidence in a court case.

Common questions

Why not just blur? Because a blurred field looks like you are hiding something. Vizard Agent replaces instead.

What names should I choose? Plausible ones. Vizard Agent will use whatever you give it, including boring realistic names.

Will it match the interface font? Vizard Agent measures the field and matches the weight and size as closely as it can.

What about the autocomplete list? Vizard Agent masks those specifically — they are the commonest leak.

Can it handle names in a table? Yes. Vizard Agent measures the result rows separately from the fields.

What if a name is spoken in the audio? That is a separate fix. Tell Vizard Agent and it can handle the audio too.

Will the video look edited? Not if the replacements are checked. Vizard Agent inspects each at full resolution.

Can it replace an email address? Yes, the same way. Give Vizard Agent the substitute you want shown.

What if the text scrolls? Harder. Vizard Agent will tell you when a re-record is the better option.

Does it change the recording otherwise? No. Only the named regions are covered.

Can I use different names in different places? Yes. Tell Vizard Agent which substitute goes where.

How long does this take? Most of the time goes on measurement, and it is a good deal quicker than re-recording.

What about a number that has to stay consistent? Say so and Vizard Agent uses the same substitute everywhere it appears.

Can I get a list of what was replaced? Yes. Ask Vizard Agent for the regions and timecodes it covered.