Daniel Steiner
Download CV

Wisp Auto Assembly

Long-form production cut from 30+ hours to under 20 hours per video.

Project
Wisp
Role
Designer and builder (AI-assisted development)
Tools
Adobe Premiere Pro (UXP panel), Google Sheets, SRT subtitles, Claude, ChatGPT, ElevenLabs
Period
2026
Premiere Pro timeline for the Wisp project Cursed Objects Warren after assembly: image clips placed on tracks V2 and V3
Premiere Pro timeline for the Wisp project Cursed Objects Warren before assembly: narration audio, subtitles, background and section reference only
Before and after assembly, captured on the Wisp project “Cursed Objects Warren”.

At a glanceProducing a complete long-form video went from more than 30 hours to under 20 hours. The system is still being developed to cut that further.

The panel assembling the same sequence, in real time.

Context & objective

Objective

Stop rebuilding the same editing moves by hand for every video, while keeping every creative decision under my control.

Challenge

A 12-minute Wisp video uses dozens of images. Each needs timing to the narration, a layout, a slot, an entrance, an exit and a transition. By hand, every video repeats the same moves.

Creative strategy

Separate decisions from execution. Decisions live in a scene plan, one row per image. Execution lives in a Premiere Pro panel.

  1. SheetStep 3
  2. TXTStep 4
  3. PreflightStep 5
  4. AssembleStep 6
  5. OverlaysStep 7

Production process

  1. Freeze the script; produce the ElevenLabs narration and its SRT.
  2. Correct the SRT against the script: the script is the text authority, the SRT the timing authority. Timestamps are never invented; conflicts stop the run.
  3. Fill the scene plan in Google Sheets with dropdowns: section, layout, image, slot, start phrase, end phrase, entrance, exit.

Step 3

  1. Convert the sheet into a structured Scene Plan TXT file with an AI-assisted step.
  2. Run preflight validation on the project, sequence and plan.
  3. Assemble: the panel finds each start phrase in the SRT, places each image on its track in the chosen layout and slot, and applies entrances, exits and Film Dissolves.
  1. Add section overlays: the section reference and logo mark fade in and out with each section.
  2. Fix individual clips with one-click entrance and exit animations.

Steps 5, 7 and 8

Reusable layouts: Rectangle, Horizontal, Trio, Vertical + Horizontal.

Track structure: background, scene visuals, section references, branding, captions.

Deliverables & contribution

Deliverables

Used in production on Every Starseed Explained in 13 Minutes (13:23).

Contribution

I designed the workflow, production logic, automation rules, inputs and outputs, and creative constraints. I built it with AI-assisted development: Claude and ChatGPT translated my specifications into code and helped me troubleshoot and iterate.

Result

Producing a complete long-form video went from more than 30 hours to under 20 hours. The system is still being developed to cut that further.

Learnings

The biggest gain came from defining creative decisions before entering the timeline. Automation works far better when layouts, timing rules, transitions and exceptions are explicit rather than ambiguous. Next: reduce the remaining manual cleanup and expand what the system automates reliably.