Perplexity Computer: From Prompt to Finished Work
A quiet UI-led launch film turns complex requests into visible agent work, artifacts, and integrations.
Field notes / why it works
The Ask/Computer switch explains the new mode by 0:04, and a concrete request at 0:08 becomes visible subagent work at 0:12 and deliverables by 0:22–0:40. A second task changes into a four-step execution stack at 0:47, the strongest replay point, before a third use case and integrations broaden the claim at 0:58–1:15.
Plate 01 / The format in one breath
A restrained, mostly wordless launch film introduces an assistant by moving from a simple input control to increasingly complex, visible work. It shows three concrete jobs and their outputs, then compresses the wider capability set into a short montage and logo finish. The promise begins with the product name at 0:00.5 and becomes tangible when the first detailed request appears at 0:08–0:11.
Why it works
- At 0:02 the basic Ask/Computer switch gives the viewer a one-gesture explanation of the new mode before any feature list.
- The long finance task at 0:08–0:11 is followed by a “Running Subagents” state at 0:12.6, analysis and spreadsheet outputs by 0:22–0:39. This makes an abstract automation claim visible.
- The second task peaks in replay at 0:46.5–0:47.5: a text request to make a short clip turns into a four-step execution stack. A new request at 0:58 then proves the pattern generalizes.
Plate 02 / Format card
| Reference | |
|---|---|
| Platform · aspect · length | YouTube · 16:9 · 640×360 source · 94.0 s at 30 fps |
| Pace | 2 measured shots, one definite cut at 1:33; 0.6–2.6 cuts/min allowing uncertain changes; 26 internal steps; 17.2 picture changes/min; measured average shot 47.01 s. The apparent edit rhythm comes from UI-state changes, not conventional cuts. |
| Script | About 18 detected spoken words, almost all around 0:50–0:55; 25 wpm across the full film, about 164 wpm during speech. Silent reading of short UI text carries most of the narrative. |
| Voice | One brief source-video/interview voice; no continuous narrator. The automatic pitch and transcript are unreliable over music and embedded speech. |
| On screen | Approximately three quarters UI demo/motion, one fifth title/brand/capability cards, and five seconds of vertical human footage (visual estimate). |
| Captions | No running narrator captions. The vertical insert has embedded all-caps, white/yellow subtitles near its bottom. Otherwise text is native to UI cards and simple editorial titles. |
| Sound | Continuous tonal, music-like bed; roughly 107 BPM is a low-confidence tempo heuristic. Integrated −15.1 LUFS, true peak −2.7 dBFS. |
| Structure | Name → mode switch → complex task → results → second task → third task and integrations → capability montage → logo. No spoken CTA; end card at ~1:31. |
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