Generate Short-form Clips from YouTube Videos with GPT-4o, Grok & Airtable

Go to Workflow
0 views
Built by Jordan Jordan
Created on June 07, 2026

Description

This n8n template demonstrates how to automate YouTube content repurposing using AI. Upload a video to Google Drive and automatically generate transcriptions, A/B testable titles, AI thumbnails, short-form clips with captions, and YouTube descriptions with chapter timestamps.

Use cases include: Content creators who publish 1-2 long-form videos per week and need to extract 5-10 short-form clips, YouTube agencies managing multiple channels, or automation consultants building content systems for clients.

Good to know

Processing time is approximately 10-15 minutes per video depending on length
Cost per video is roughly $1.00 (transcription $0.65, AI generation $0.35)
YouTube captions take 10-60 minutes to generate after upload - the workflow includes automatic polling to check when captions are ready
Manual steps still required: video clipping (using provided timestamps), social media posting, and YouTube A/B test setup

How it works

When a video is uploaded to Google Drive, the workflow automatically triggers and creates an Airtable record
The video URL is sent to AssemblyAI (via Apify) for transcription with H:MM:SS.mmm timestamps
GPT-4o-mini analyzes the transcript and generates 3 title variations optimized for A/B testing
When you click "Generate thumbnail" in Airtable, your prompt is optimized and sent to Kie.ai's Nano Banana Pro model with 2 reference images for consistent branding
After uploading to YouTube, the workflow polls YouTube's API every 5 minutes to check if auto-generated captions are ready
Once captions are available, click "Generate clips" and Grok 4.1 Fast analyzes the transcript to identify 3-8 elite clips (45+ seconds each) with proper start/end boundaries and action-oriented captions
GPT-4o-mini generates a YouTube description with chapter timestamps based on the transcript
All outputs are saved to Airtable: titles, thumbnail, clip timestamps with captions, and description

How to use

Duplicate the provided Airtable base template and connect it to your n8n instance
Create a Google Drive folder for uploading edited videos
After activating the workflow, copy webhook URLs and paste them into Airtable button formulas and automations
Upload your edited video to the designated Google Drive folder to trigger the system
The workflow automatically generates titles and begins transcription
Add your thumbnail prompt and 2 reference images to Airtable, then click "Generate thumbnail"
Upload the video to YouTube as unlisted, paste the video ID into Airtable, and check the box to trigger clip generation
Use the provided timestamps to manually clip videos in your editor
Copy titles, thumbnail, clips, and description from Airtable to publish across platforms

Requirements

Airtable account (Pro plan recommended for automations)
Google Drive for video upload monitoring
Apify account for video transcription via AssemblyAI actor
OpenAI API key for title and description generation (GPT-4o-mini)
OpenRouter API key for clip identification (Grok 4.1 Fast)
Kie.ai account for AI thumbnail generation (Nano Banana Pro model)
YouTube Data API credentials for caption polling

Customising this workflow

Tailor the system prompts to your content niche by asking Claude to adjust them without changing the core structure
Modify the clip identification criteria (length, caption style, number of clips) in the Grok prompt
Adjust thumbnail generation style by updating the image prompt optimizer
Add custom fields to Airtable for tracking performance metrics or additional metadata
Integrate with additional platforms like TikTok or Instagram APIs for automated posting

Nodes Used (8)

Airtable
n8n-nodes-base.airtable
Basic LLM Chain
@n8n/n8n-nodes-langchain.chainLlm
Code
n8n-nodes-base.code
Google Drive
n8n-nodes-base.googleDrive
HTTP Request
n8n-nodes-base.httpRequest
OpenAI
@n8n/n8n-nodes-langchain.openAi
OpenRouter Chat Model
@n8n/n8n-nodes-langchain.lmChatOpenRouter
YouTube
n8n-nodes-base.youTube