Edit Your Videos With ChatGPT

Use ChatGPT to script cuts, captions, and edit decisions for faster video turnaround.

Most video creators waste hours on edit decisions that have nothing to do with creativity — figuring out where to cut, what B-roll to call out, how to write captions that match the pacing. ChatGPT can't open your timeline, but it can do the thinking before you touch your editor. You give it your script or transcript, it gives you a cut-by-cut breakdown, caption text, and a clear editing roadmap. The result: you spend less time staring at a blank timeline and more time actually finishing videos.

This works best for talking-head content, educational videos, short-form reels, and podcast clips — anything where you have a spoken script or a rough transcript to start from. If you don't have a transcript yet, run your raw audio through a tool like Whisper or Descript first, then bring that text into ChatGPT.

The Prompt

Paste this into ChatGPT (GPT-4 or later). Replace the bracketed sections with your actual content.

You are a professional video editor helping me plan the edit for a [SHORT-FORM / LONG-FORM] video.

Here is the full script / transcript:
[PASTE YOUR SCRIPT OR TRANSCRIPT HERE]

Please do the following:

1. CUT LIST — Go through the script line by line and flag:
   - Any filler words, repeated phrases, or dead air to cut
   - Sections that drag or could be tightened
   - Natural pause points where a cut or transition would feel clean
   Label each flag with a timestamp range if I gave you one, or with a quote from the transcript.

2. B-ROLL CALLOUTS — List 5–8 specific moments where B-roll would strengthen the video. For each one, write:
   - The line from the script it accompanies
   - A concrete description of what the B-roll shot should show

3. CAPTION LINES — Rewrite the first 30 seconds of the script as formatted caption lines. Each line should be 4–6 words max, punchy, and match natural speech rhythm. Bold any high-emphasis words.

4. HOOK RATING — Rate the first 5 seconds on a scale of 1–10 for viewer retention. Give me one specific rewrite if the rating is below 8.

5. EDIT PACING NOTE — In 2–3 sentences, describe the overall pacing this video should have (fast cuts, breathing room, energetic vs. calm) based on the content and likely audience.

How to Use the Output

When ChatGPT returns the breakdown, don't just read it — open your editor side-by-side and work through each section in order. The cut list becomes your first rough cut guide. The B-roll callouts become your shot pull list. The caption lines go straight into your caption tool or subtitle file. You're not replacing judgment — you're front-loading the cognitive work before you're buried in the timeline.

For long-form videos (10+ minutes), break the transcript into chunks of 3–5 minutes and run the prompt multiple times. Trying to process a 45-minute transcript in one shot degrades the output quality.

Tips for Better Results

Include timestamps in your transcript if you have them — Descript, Otter.ai, and Whisper all output timestamps by default. This makes the cut list immediately actionable in your editor.
Tell ChatGPT your platform. A YouTube tutorial and a 30-second Instagram reel have completely different pacing requirements. Add 'This is for [platform]' at the top of the prompt.
Specify your audience in one sentence. 'This is for coaches who are new to AI tools' will produce different B-roll callouts than 'This is for experienced video editors.' Context changes the recommendations.
If the hook rating comes back low, ask ChatGPT to give you three alternative opening lines, not just one. Then pick the one that sounds most like you.
Use the caption output as a starting draft, not final copy. Read it out loud against your footage — anything that feels off, fix it manually. ChatGPT doesn't know your exact inflection.
For talking-head videos, ask ChatGPT to flag every sentence longer than 15 words. Long sentences in video scripts almost always need a cut or a breath break inserted.
Save your best-performing prompt setup as a ChatGPT custom instruction or a saved snippet in a notes app. You want to run the same structure every time so your edits stay consistent across videos.

Variations

Once you have the core prompt working, these variations handle specific editing problems:

Variation 1: Clip Extraction for Short-Form

Here is a transcript from a [X-minute] long-form video:
[PASTE TRANSCRIPT]

Identify the 3 best standalone clips for short-form content (Instagram Reels, TikTok, YouTube Shorts). For each clip:
- Quote the exact start and end lines from the transcript
- Explain in one sentence why this section works as a standalone clip
- Write a short-form caption (under 100 characters) for it
- Suggest one on-screen text overlay for the hook frame

Variation 2: Thumbnail and Title Extraction

Here is the script for my video:
[PASTE SCRIPT]

Give me:
1. Five thumbnail concepts — each described as a visual scene (what's in the frame, any text overlay, the emotion on my face)
2. Five YouTube title options — specific, curiosity-driving, no clickbait, under 60 characters
3. The single most quotable line from the script that could work as a standalone social post

Variation 3: Reformat a Podcast Episode into a Script

Here is a rough transcript from a podcast conversation:
[PASTE TRANSCRIPT]

I want to turn this into a structured solo video script. Please:
1. Remove all back-and-forth dialogue — keep only the content that works as a direct-to-camera monologue
2. Rewrite it in first person, as if I'm speaking directly to a viewer, not another host
3. Tighten sentences to under 15 words where possible
4. Add three natural transition phrases between sections so the edit doesn't feel abrupt
5. Flag any sections that reference 'you' (the other host) that need to be reframed for a solo format

What This Doesn't Replace

ChatGPT doesn't watch your footage. It doesn't know if your lighting changed mid-shot, if you fumbled a word but your face was perfect, or if the energy in your voice dropped in a section that looks fine on paper. The output is a plan, not a finished edit. You still make the final call on every cut. What changes is how long it takes you to get there — and how much mental energy you have left once you sit down at the timeline.

Start with one video you've already edited. Run the prompt on that script and compare ChatGPT's cut recommendations to the choices you already made. You'll immediately see where it's useful and where you'd override it — that calibration is more valuable than any tutorial.