WEBVTT

NOTE
Captions for the introduction video, carrying only what the cut does not
already put on screen.

The narration has five lines. Two of them — "Someone is always on duty" at
0:22 and "XInfer.AI is the merchant's AI agent" at 1:05 — are rendered in the
video itself (src/scenes/Section2.tsx in the intro-video repo). Captioning
those would print the same words twice in the same frame, so they are omitted
deliberately rather than missed.

What remains is the problem setup, which is audio-only, and is exactly the
part a muted autoplay would otherwise lose.

The agent's own speech through the demo is not captioned. She is full-screen
in the app with no chat text on display, so there is no frame to read her
words from, and captions written from memory would be captions that lie. The
customer's question is captioned because it is recorded verbatim in the
intro-video repo's docs/YOUTUBE.md.

Re-time these if the cut changes.

1
00:00:05.000 --> 00:00:11.500
It's Friday night. Your store is open.
A customer has a question.

2
00:00:12.000 --> 00:00:15.500
Nobody is there to answer it.

3
00:00:16.000 --> 00:00:20.000
He isn't the only one.

4
00:00:39.000 --> 00:00:43.000
I'm looking for a gift under 20.