Watch the Agent’s Work, Then Judge It
A two-person product demo replaces an abstract feature claim with task footage, a visible failure, and a better result.
Field notes / why it works
A question at 0:00 gets a plain capability answer by 0:08, with the interface visible by about 0:05. The demo spends 0:26–0:46 showing an actual result, then lets viewers compare a visibly flawed output at 1:11–1:19 with a better one at 1:22–1:25; both comparison moments align with replay peaks. The closing 1:26–1:40 translates the feature into a higher-level decision the user can make.
Plate 01 / The format in one breath
Two colleagues at a desk introduce a capability through a short question and answer, then show the actual software carrying out a concrete task. The viewer watches a result, sees a less successful comparison, and hears what this changes about the way the team works.
Why it works
- The promise arrives by 0:08: the agent can test its own work and return a video; the screen recording starts by about 0:05.
- The proof is layered: a real task is set up at 0:12–0:26, a working recording is evaluated at 0:26–0:46, and a second example starts near 1:00.
- The visibly flashing failed example at 1:11–1:19 creates a genuine judgment moment before a better result at 1:22–1:25. The 1:13–1:14 and 1:22–1:23 replay peaks coincide with that comparison.
Plate 02 / Format card
| Platform · aspect · length | YouTube · 16:9 · 1280×720 · 23.98 fps · 102 s |
| Pace | 44 detected shots · 43 detected cuts / 25.3 per minute · 57 picture changes / 32.9 per minute · 2.32 s average shot. The flashing screen from 1:11–1:19 accounts for 16 cuts in one ten-second window; ordinary conversational passages are much slower. |
| Script | 357 spoken words · 215 wpm across the film, 249 wpm while speaking · 23 sentences averaging 15.5 words · two questions. |
| Voice | Two low-register, quick, conversational adult voices: a curious host and a technically fluent demonstrator. |
| On screen | Approx. 55% studio conversation, 45% laptop or desktop proof (inferred from shot durations and sheets). |
| Captions | No burned speech captions; the only large type is the thumbnail title. Product UI text remains native to the recording. |
| Sound | Dry dialogue and screen audio/room tone; no audible music bed or designed cut effects. Measured −25.3 LUFS, true peak −3.4 dBFS. |
| Structure | Opening question → capability → real task → proof → comparison → workflow takeaway. No spoken CTA or 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