Edit Videos with GPT-6 Astra Inside Premiere Pro

Technology / Workflow Overview

Chat Video Pro is presented as a tool that runs inside Adobe Premiere Pro, using a Story Cutter Assistant workflow to convert long-form source content (e.g., a 3-hour podcast) into:

• Rough cuts
• Standalone short-form YouTube clips

The workflow uses LLM “models” to:

1. Plan the edit (via a creative brief / story plan).
2. Execute the actual cut in the timeline by creating micro-clips, hooks, value sections, and payoffs.

---

How Astra (GPT-6 Astra / “Astra”) Is Used

Astra is the “architect/big brain” model used early to understand the content and produce a creative brief.

In the demo:

• The creator uploads a transcript.
• Astra generates a fine-tuned creative brief including structure elements such as:
  • Angle/goal
  • Suggested video format
  • A rough ~3 minute behind-the-scenes explainer
  • Beats of the video: hook → value → payoff

After the plan is generated, the editor can jump to suggestions via timestamped “paper cut” soundbites that link directly back to the source footage.

---

Transcript Ingestion Options (Two Paths)

1. Free / Premiere-native transcript generation

• Use Premiere Pro’s text panel → export transcript

2. Higher-fidelity option inside Chat Video Pro

• Upload a transcript using “transcribe with Storycribe”
• Ingest the transcript into Story Cutter and add a creative brief

---

Rough Cut to Timeline Features

The assistant provides timeline support such as:

• Markers for hook/value sections at the top of the timeline
• Micro clips derived from the long recording
• “Insert to timeline” to add recommended segments one at a time for manual story-building
• Model recommendations for edit improvements like:
  • Adding B-roll / visual breaks
  • Improving pacing
• An “insert rough cut” action that automates rough-cut generation and continues the cut across the timeline

---

Cost / Model Strategy (Analysis & Guidance)

Astra is described as very expensive, with example pricing discussed in terms of tokens.

Suggested strategy:

• Use Astra only for planning (creative brief)
• Switch to cheaper execution models for the actual editing

The demo recommends Luna / Terra as mid-cost “project manager” models for execution after planning.

Optimization for scripted workflows

• If a script is already provided and the output should be verbatim, it’s unnecessary to use Astra for planning.
• Instead:
  • Use Astra for setup/verification
  • Then switch to GPT 5.6 Terra (faster/lower cost) because the task is more binary/deterministic
• The system performs script-match edits and supports batch editing.

---

Scripted Batch-Edit Tutorial (Multiple Outputs)

Example use case

• The creator has 5 scripts for 5 videos
• Uploads scripts (TXT/Markdown) in the creative brief

System behavior

• Identifies script matches and lists what will be edited
• Performs batch editing by spawning 5 sub-agents (one per scripted video)
• After completion:
  • All edited cuts are inserted
  • They can be reviewed individually or together
• Segments are linked to the correct sequence from the source

---

Product Value Claims / Use-Case Positioning

Key emphasis: reducing analysis paralysis.

Instead of watching 3 hours of footage, the editor can:

• Iterate quickly using AI-generated rough cuts

Mentioned suitability includes:

• AB testing hooks
• Finding best engaging moments
• Rapid consolidation of raw/podcast content into short-form YouTube clips

Export guidance

• If no changes are needed, the editor can:
  • Close gaps
  • Export a ready ~3 minute social/podcast clip

---

Main Speakers / Sources

• The primary speaker/source is the “Chat Video Pro” creator demonstrating the workflow (including references to Joe Rogan / MrBeast-style example content).
• No other explicit named interviewer appears beyond mentions of Joe Rogan and Mr. Beast as example content subjects.