How It’s Done

Visual process evidence

← Back to interview

WT-000006 — Narrated screen walkthrough

Linked to Process-mapping interview INT-000018 · Status: recorded

Before recording: Choose only the task window, avoid private information, and speak aloud what you're doing, what you're checking, and why. The recording is saved for review. AI video analysis is optional and sends the complete fictional recording to Gemini only after you click Analyze full video. SOP changes always require human approval.

Keep the demonstration broad at first. Knowledge Discovery can ask about a section only after you establish whether it belongs to this employee.

1. Saved recording

Recorded 2026-11-18T17:59:29+00:00 · Reported duration: 01:07 · Stored in your application data directory. You control whether to send the recording to Gemini for action analysis.

Legacy fallback: fixed-interval screenshots every 10 seconds. Smart video analysis below suggests timestamps based on actual actions instead.

2. Spoken narration and reviewer notes

Automatic speech-to-text is attempted during every recording. Check the saved transcript for accuracy. If recognition was unavailable, enter or paste a transcript before running AI analysis; the video alone cannot supply reliable narration text.

3. Action-aware screenshot suggestions (4)

Gemini analyzes the actual screen recording and narration, identifies meaningful actions, then recommends the moment to capture each screenshot. The browser saves the selected frames locally. Results are unverified until you review them. Approving a suggestion does not attach it to an SOP; use the SOP step attachment tools below for that.

A Gemini key is required. Connect Vision AI first.

Opt-in: sends the complete WebM (including audio), corrected transcript, caption timestamps, and limited role/SOP context to Gemini. The Gemini file is deleted after analysis. Free-tier limits apply. Fictional demonstrations only.

Select Analyze full video to identify actions and capture candidate screenshots.

Enter recipient address · 00:15

Selected recipient email address in the 'To' field

Step linkage not verified · Evidence confidence: high

AI-suggested candidate screenshot

The user selects the recipient from the dropdown list to address the email correctly.

Narration: So first what I do is I select who I actually want to send this email to and I put it right here.

Enter subject line · 00:36

Typed the subject line for the email

Step linkage not verified · Evidence confidence: high

AI-suggested candidate screenshot

The user fills in the subject field describing the course and topic.

Narration: Then I add a title. So for mine I include the subject which is the course that I'm taking marketing.

Compose email content · 00:56

Entered the main body text of the email

Step linkage not verified · Evidence confidence: high

AI-suggested candidate screenshot

The user inputs the opening, main content, and closing remarks in the email body.

Narration: And then I actually go in and I type the content of the email... whatever I want to add to the email I generally put it at the end.

Send email · 01:04

Clicked the 'Send' button

Step linkage not verified · Evidence confidence: high

AI-suggested candidate screenshot

The user completes the process by clicking send after verifying all details.

Narration: once the email is done I have my content, title, who it's sending to, I just click send and I'm good from there.

4. Saved screenshot library (11)

Candidate frames are samples, not confirmed illustrations. Avoid attaching repeats or frames with irrelevant content. Once attached, they appear in approved SOP Word exports.

00:02 · Not attached Candidate frame at 00:02

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

00:12 · Not attached Candidate frame at 00:12

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

00:15 · Not attached Candidate frame at 00:15

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

00:22 · Not attached Candidate frame at 00:22

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

00:32 · Not attached Candidate frame at 00:32

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

00:36 · Not attached Candidate frame at 00:36

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

00:42 · Not attached Candidate frame at 00:42

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

00:52 · Not attached Candidate frame at 00:52

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

00:56 · Not attached Candidate frame at 00:56

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

01:02 · Not attached Candidate frame at 01:02

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

01:04 · Not attached Candidate frame at 01:04

This interview demo remains reference evidence. Step attachment becomes available from an approved SOP walkthrough.

AI-assisted walkthrough analysis

Opt in to analyze up to six saved video screenshots together with narration and linked SOP information. This separate optional analysis uses only saved screenshots (up to six); it does not send the full video. Review each suggested question before launching Knowledge Discovery.

To analyze screenshots, first connect your Gemini Vision AI key.

Narration transcript saved and ready for review-based analysis.