When an AI Invents the Brand Story
A polished two-person comedy ad turns unreliable AI answers into a sourcing problem, then resolves it with a brand-owned answer.
Field notes / why it works
A mundane customer question is clear by 0:08, while the assistant's first absurdly specific claim lands in the most-replayed 0:17.9–0:18.8 window. The source challenge at 0:31 makes the brand problem concrete, and a better sourced answer at 1:06 earns the final call to action. Twenty-eight shots and 19 in-shot changes keep a 94-second dialogue sketch visually active.
Plate 01 / The format in one breath
A customer interviews a personified, overconfident AI about an ordinary purchase question. Its wrong answers get more specific and absurd until the customer challenges the source; a verifiable product answer then resolves the problem. The brand's promise is saved for the final six seconds, after the sketch has demonstrated why the viewer needs it.
Why it works
- The branded logo opening turns into a face-to-face comic premise by 0:04.6, with the customer's actual question at 0:07.9. The viewer has a situation to follow before the technical explanation arrives.
- Escalation combines dialogue and evidence-like graphics: a bizarre claim at 0:17.7, a dictionary-style card around 0:24.8, a dubious review card around 0:55, then a grounded alternative at 1:05.6. The peak replay window, 0:17.9–0:18.8, lands on the first absurd product claim; replay alone does not prove why people returned.
- The brand argument is earned by the plot: the assistant admits at 0:32.1 that missing site information forced it to assemble an answer elsewhere, and the better sourced answer begins at 1:05.6.
Plate 02 / Format card
| Platform · aspect · length | YouTube · 16:9 · 94.0 s · measured 640×360, 23.98 fps |
| Pace | 28 shots; 27 definite cuts, or 17.2–17.9 cuts/min allowing one uncertain change; 29.4 picture changes/min including 19 in-shot steps; mean shot 3.36 s, median 2.54 s |
| Script | 245 spoken words; 165 wpm over the full runtime, about 215 wpm during speech; 32 short sentences averaging 7.7 words; five questions |
| Voice | Two adult conversational archetypes: a politely impatient customer and a genial, overconfident assistant; the measured medium-register, very animated voice centers near 148 Hz |
| On screen | Almost all live-action two-person wide shots and alternating medium close-ups; one ~2 s musical cutaway and a few restrained UI cards (inferred mix: ~95% live action) |
| Captions | No burned speech subtitles; sparse fictional answer and review cards appear behind or beside the assistant |
| Sound | Continuous, low tonal bed measured around −32.3 dBFS between words; dialogue forward; −19.4 LUFS integrated, −1.2 dBFS true peak |
| Structure | Silent branded reveal → customer question → wrong answer → source challenge → comic escalation → grounded answer → final joke → brand line and 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 4 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