ChatGPT Work: from request to finished deck
A quiet 100-second agent launch film proves its promise through one spoken task, cross-app outputs, a revised deck, and scheduled follow-through.
Field notes / why it works
The white-space logo opening delays the promise until about 0:05, then a single spoken request at 0:14–0:20 gives the demo a clear cause. Each next screen adds tangible proof: supplier table by 0:37, imagery by 0:47, deck by 0:58, chart revision by 1:11, and scheduled follow-up by 1:25. Twelve main shots and 20 internal changes keep a patient 100-second film moving without dense narration.
Plate 01 / The format in one breath
A minimalist launch film introduces an agent, lets a human give it one ambitious spoken assignment, then follows that assignment across apps until a polished deliverable and an ongoing follow-up appear. The engine is visible cause and effect: request → research → creation → revision → scheduled action.
Why it works
- A tiny logo grows in an empty white field for the first 3 seconds; the product promise arrives as large, centered type around 0:05–0:08. The empty start makes that reveal feel substantial.
- The single spoken command at 0:14–0:20 names three workstreams. The next minute shows those workstreams as a supplier search, spreadsheet, product imagery, slides, and chart revision instead of listing capabilities.
- The payoff keeps escalating: a downloadable deck appears at 0:57–0:59, a slide gets changed at 1:05–1:11, and a phone shows scheduled work and later feedback at 1:16–1:29.
Plate 02 / Format card
| Reference | |
|---|---|
| Platform · aspect · length | X · 16:9 · 1280×720 · 60 fps · 100 s |
| Pace | 12 detected shots, 11 cuts, 6.6 cuts/min; 20 changes inside shots, about 18.6 total picture changes/min; mean shot 8.33 s, median 4.24 s. Long opening and ending contain multiple internal changes. |
| Script | 27 transcribed spoken words, 19 wpm across the whole film; about 104 wpm during the short spoken request. Later isolated punctuation detections are not usable speech. |
| Voice | One medium-register adult requester, conversational and purposeful. No continuous narrator. |
| On screen | About 20 s logo/typography/command UI; about 55 s desktop app, artifact and product imagery; about 15 s phone and message views; about 10 s closing type (approximate from sheets). |
| Captions | No speech subtitles. Centered headline cards and legible native UI text carry the story. |
| Sound | Continuous tonal, music-like bed measured by the tool; roughly 99 BPM estimate. No reliable repeated cut-hit pattern. −12.9 LUFS integrated, +0.1 dBFS measured true peak. |
| Structure | Brand/promise → command → cross-app work → finished output → revision/follow-through → end card. |
Pro / 13 plates + recipe
The first plates are open. The full dissection is Pro.
Get the timed shots, type and sound specs, and the runnable recipe.
Open the full blueprint- Plate 03 / The hook, frame by frame (first 3 seconds)Pro
- Plate 04 / Structure (beat sheet)Pro
- Plate 05 / Shot grammarPro
- Plate 06 / Visual style guidePro
- Plate 07 / Voice and deliveryPro
- Plate 08 / Sound designPro
- Plate 09 / PackagingPro
- Plate 10 / Script templatePro
- Plate 11 / Production planPro
- Plate 12 / PromptsPro
- Plate 13 / Edit recipePro
- Plate 14 / QA checklistPro
- Plate 15 / Make it yoursPro
- Recipe / runnable promptPro