Turn YouTube Comments into Content Ideas with GPT-4.1-mini, Tavily & Apify

Description

This n8n template demonstrates how to turn raw YouTube comments into research-backed content ideas complete with hooks and outlines.

Use cases include:

Quickly mining a competitor’s audience for video ideas.
Generating hooks and outlines for your own channel’s comments.
Validating content opportunities with live audience feedback.

Good to know

Apify is used to scrape YouTube comments (requires an API token).
GPT-4.1-mini is used for both filtering and content generation.
Tavily provides fresh research to ground the AI’s responses.
All outputs are stored in Google Sheets, making it easy to manage and track ideas.

How it works

Trigger – Paste a YouTube URL into the chat trigger.
Scrape Comments – Apify fetches all comments and metadata.
Filter – GPT-4.1-mini decides if each comment could inspire a content idea.
Store – Comments and “Yes/No” decisions are appended to Google Sheets.
Research & Enrich – For “Yes” comments, Tavily provides context, and GPT generates a topic, hook, and outline.
Update Sheet – The same row in Google Sheets is updated with enriched fields.

Google Sheets Setup

Your Google Sheet should include these columns (in this order):

id | text | author | likes | isIdea | topic | research | hook | outline id** – unique identifier for each comment
text** – the full YouTube comment
author** – commenter’s name/handle
likes** – number of likes on the comment
isIdea** – “Yes” or “No” depending on GPT filter
topic** – extracted video topic
research** – 300–500 word background from Tavily
hook** – engaging opening sentence for a video
outline** – structured video outline

Setup Steps

Connect your Apify, OpenAI, Tavily, and Google Sheets credentials in n8n.
Point the Google Sheets nodes to your own document and ensure the above headers exist.
Replace sample API keys with your own stored in n8n Credentials.

Time to set up: ~15–25 minutes for a first-time n8n user (less if you already have credentials handy).

Customizing this workflow

Filter logic** – Loosen the GPT filter to allow borderline ideas, or tighten it to only accept the best ones.
Research depth** – Change Tavily’s search depth (e.g., depth: basic vs depth: advanced) to control how detailed the background research is.
Notification channels* – Send new “Yes” ideas directly to Slack (#content-ideas), Notion (your content board), or Email* (notify the content manager instantly).
Alternative outputs** – Instead of hooks/outlines, generate:
A script draft for YouTube Shorts.
Blog post angles based on the same audience comments.
A poll question for community engagement.