Sensitive Data
PII Auto-Redact™
PII Auto-Redact™ scans a recording for personal data and hides it before the video leaves your computer. The scan runs in the Enterprise Desktop App, and the app re-encodes the video with every hidden match applied, so only the redacted version is uploaded. The unredacted recording never reaches Tutorial AI’s servers.
It’s built for teams that record real workflows with real customer data: banking, healthcare, insurance, the public sector. You don’t need test accounts or staged data. Record the real thing and remove what viewers shouldn’t see.
Scanning a recording
Section titled “Scanning a recording”-
Record as usual in the Enterprise Desktop App.
-
When you stop recording, the save screen offers to clean it up before upload. Click Cut or blur here.
-
Click Scan recording. The scan reads the text on screen throughout the recording and lists matches as it finds them. How long it takes depends on the length of the recording and on your computer’s performance. Long recordings can take a while, and you can keep blurring by hand while it runs.
-
Review the matches (see below), then save. The app re-encodes the video on your computer and uploads only the redacted version.
What it finds
Section titled “What it finds”- Names
- Email addresses
- Phone numbers
- Addresses
- Card and account numbers
- ID numbers
- Passwords, API keys and tokens
Reviewing the matches
Section titled “Reviewing the matches”Matches are grouped by type, with a count for each group. The top of the list shows how many are still to review, and Search matches finds a specific value.
Each match is marked by how sure the scan is:
- Likely: almost certainly personal data, outlined in red
- Possible: might be personal data, outlined in amber with a dashed line. Check these.
The same outlines appear on the video, so you can see where each match sits on screen. Each row has a small thumbnail of that part of the screen, the matched text, and a toggle with three positions:
| Toggle | Meaning |
|---|---|
| Dot in the middle, red or amber ring | To review: not decided yet |
| Purple, dot on the right | Hidden in the uploaded video |
| Grey, dot on the left | Kept: shown as it is. The row is dimmed. |
Deciding a whole group at once
Section titled “Deciding a whole group at once”Click Hide all next to a group to hide every match in it that you haven’t kept. Use it for types you never want to show, such as email addresses or API keys, and decide other groups, such as names, one match at a time.
Lookalike text or blur
Section titled “Lookalike text or blur”By default, a hidden match is replaced with lookalike text: a realistic fake of the same type. A name becomes another name, an email another email. The screen still reads like the real product, with no real person in it. Matches that can’t be replaced as text, and boxes you draw yourself, are blurred instead.
To blur a single match instead, or to switch a blurred match back to lookalike text, use the ⋯ menu on its row.
The thumbnail on each row shows what viewers will see: the lookalike text, the blur, or the original if you kept it.
Scanning again
Section titled “Scanning again”Click Rescan to run the scan on the recording again.