AI YouTube Video Generator in n8n: Sheets to GPT-4o Script

Turn Google Sheets YouTube ideas into a rendered video—then publish only after approval

This n8n workflow automates your YouTube pipeline: it pulls “to create” ideas from Google Sheets, uses GPT-4o to generate a 14-scene script with voice + image prompts, renders the video via JSON2Video, writes the output URL back to Sheets, and uploads to YouTube only after a manual “approved” step.

What this workflow does

• Scheduled intake: Runs on a schedule and fetches the first Google Sheets row where PRODUCTION STATUS is set to “to create”.
• GPT-4o scripting (JSON output): Sends the idea title, captions, and style fields to OpenAI (GPT-4o) to generate a 14-scene JSON script with two-sentence voiceover text and cinematic image prompts.
• Music/intro selection from Sheets: Reads the selected “final” music/intro assets from a second Google Sheets tab.
• Video rendering with JSON2Video: Submits the script, voice settings (via ElevenLabs), image model, and audio/intro URLs to the JSON2Video API to start rendering.
• Polling + status handling: Waits, polls render status, and repeats until the render completes or errors out.
• Writes results to Sheets: On completion, saves the JSON2Video movie URL back to Google Sheets and marks the row as created; on failure, logs an error status.
• Manual approval → YouTube publishing: When PUBLISHING STATUS is set to “approved”, it downloads the finished video from the stored URL, uploads it to YouTube, applies privacy/compliance status via the YouTube Data API, and marks the row as published.

Use cases

• SaaS teams producing repeatable YouTube content from a backlog managed in Google Sheets
• Automation engineers standardizing a GPT-4o → JSON2Video render workflow with controlled approvals
• Operators who want “draft-ready” rendering tracked in Sheets before any YouTube upload

Technical details

• Google Sheets (OAuth2) for intake, music/intro selection, and status updates
• OpenAI (GPT-4o) for generating the 14-scene JSON script
• ElevenLabs voice settings used in the JSON2Video request
• JSON2Video API via HTTP Request using an API key auth credential
• YouTube OAuth2 + YouTube Data API for upload and privacy/compliance settings
• Core n8n logic nodes include: if, wait, switch, HTTP Request, and YouTube integration nodes
• Supplemented with workflow notes (Sticky Note) for clarity