YouTube video description and SEO metadata from transcripts

Introduction

Creating optimised YouTube descriptions and SEO metadata manually is a task that eats away at your time without adding much creative value. You record or source video content, but then you're stuck transcribing, summarising, and writing descriptions from scratch. Each video that goes live without proper metadata is a missed opportunity for discovery, and each description you write by hand is time that could go elsewhere.

The good news is that this workflow can be completely automated. By combining three focused AI tools, you can take a video transcript and produce a polished description, SEO-friendly title variants, and metadata in minutes. No manual handoff, no copy-pasting between tabs, no waiting for a team member to do the work.

This Alchemy guide shows you how to wire together Copy.AI, Quick Creator, and Resoomer.AI with an orchestration tool to turn raw transcripts into publication-ready content. Whether you're running a YouTube channel, managing content for clients, or publishing multiple videos weekly, this workflow removes the friction from your publishing pipeline.

The Automated Workflow

We'll build this workflow using Make (formerly Integromat) as the orchestration engine, though the same logic applies to Zapier or n8n. Make is particularly good here because it has native integrations with most of these tools and gives you visual workflow design without needing code.

The workflow follows this path: receive a transcript, summarise it with Resoomer.AI, generate multiple title options with Copy.AI, create a full description with Quick Creator, then compile everything into a structured output you can paste directly into YouTube.

Step 1:

Trigger the Workflow

Your entry point is a transcript. You can feed this in several ways: paste it into a form, upload a file, or trigger via webhook. For this example, we'll assume you're passing a transcript via a webhook.

Copy
POST /webhook/youtube-workflow
Content-Type: application/json

{
  "videoTitle": "How to Use AI for Content Creation",
  "transcript": "So today we're going to talk about AI tools... [full transcript text]",
  "channelName": "My Content Channel",
  "targetAudience": "Content creators and marketers"
}

In Make, set up a custom webhook module to listen for this payload. The webhook URL will be something like https://hook.make.com/xxxxx. You'll reference the transcript field in subsequent steps.

Step 2:

Summarise the Transcript with Resoomer.AI

Resoomer.AI specialises in creating concise summaries. Rather than feeding the full transcript into every subsequent tool, which wastes API calls and tokens, you summarise once and reuse that summary.

Resoomer's API accepts text and returns a condensed version. Here's how to call it in your Make workflow:

Copy
POST https://api.resoomer.com/summarize
Content-Type: application/json
Authorization: Bearer YOUR_RESOOMER_API_KEY

{
  "text": "{{ 1.transcript }}",
  "length": "medium",
  "language": "en"
}

The {{ 1.transcript }} syntax references the transcript from your webhook trigger (module 1). Set the length parameter to "medium" to get something substantial but focused, roughly 150-200 words.

Store the summary output in a variable called summaryText. This becomes your source material for the description and title generation steps.

Step 3:

Generate Title Variants with Copy.AI

Copy.AI is purpose-built for generating marketing copy, which includes compelling titles. You'll ask it to create 5-7 title options that are YouTube-friendly, SEO-conscious, and click-worthy.

Set up a Copy.AI API call like this:

Copy
POST https://api.copy.ai/v1/generate
Content-Type: application/json
Authorization: Bearer YOUR_COPY_AI_KEY

{
  "prompt": "Based on this video summary, generate 7 YouTube video titles. Each title should be between 40-60 characters, include a primary keyword, and encourage clicks. Summary: {{ summaryText }}",
  "model": "gpt-4",
  "temperature": 0.7,
  "max_tokens": 500
}

Copy.AI's API returns a single text block with all titles. You'll need to parse this in the next step, so use Make's text functions to split by newline and extract individual titles into an array.

In Make, add a "Text Aggregator" module to format this into usable data:

Copy// This runs in Make's JavaScript module
const titles = text.split('\n').filter(t => t.trim().length > 0);
return { titles: titles };

Step 4:

Create a Full Description with Quick Creator

Quick Creator is designed for longer-form content generation. Feed it the summary and ask for a YouTube description that balances clarity, SEO keywords, and a call-to-action.

Copy
POST https://api.quickcreator.ai/v1/content
Content-Type: application/json
Authorization: Bearer YOUR_QUICK_CREATOR_KEY

{
  "contentType": "youtubedescription",
  "summary": "{{ summaryText }}",
  "style": "professional_engaging",
  "includeTimestamps": false,
  "keywords": ["AI tools", "content creation", "automation"],
  "callToAction": "Subscribe for more content",
  "maxLength": 5000
}

Quick Creator returns a full description (under YouTube's 5,000-character limit). Store this as descriptionFull.

Step 5:

Generate SEO Metadata

Now create tags, hashtags, and a short summary excerpt for social sharing. You can do this with a second Copy.AI call:

Copy
POST https://api.copy.ai/v1/generate
Content-Type: application/json
Authorization: Bearer YOUR_COPY_AI_KEY

{
  "prompt": "Generate YouTube metadata for this content. Provide: 1) 12-15 relevant tags (one per line), 2) 5-8 hashtags, 3) a 280-character social media excerpt. Content summary: {{ summaryText }}",
  "model": "gpt-4",
  "temperature": 0.6,
  "max_tokens": 400
}

Parse this output to extract tags, hashtags, and the social excerpt into separate fields.

Step 6:

Compile and Output

In the final step, assemble everything into a structured JSON object that you can save to a file, send to a spreadsheet, or post to your YouTube management tool.

Copy
POST https://hook.make.com/output
Content-Type: application/json

{
  "originalTitle": "{{ originalVideoTitle }}",
  "titleOptions": {{ parsedTitles }},
  "descriptionFull": "{{ descriptionFull }}",
  "tags": {{ parsedTags }},
  "hashtags": {{ parsedHashtags }},
  "socialExcerpt": "{{ socialExcerpt }}",
  "summary": "{{ summaryText }}",
  "generatedAt": "{{ now }}",
  "status": "ready_for_review"
}

Connect this to a webhook receiver (like a Google Form submission, Airtable, or a custom endpoint on your own server). If you use Make's "HTTP" module to send this, it creates a persistent record you can retrieve later.

Alternatively, save directly to Google Sheets using Make's built-in "Google Sheets" module, so your team sees the output in real time without additional setup.

Full Workflow Diagram in Make

1. Webhook Trigger (Module 1): Receive transcript
2. Resoomer.AI Request (Module 2): Summarise transcript
3. Copy.AI Request (Module 3): Generate title options
4. Text Parser (Module 4): Extract titles into array
5. Quick Creator Request (Module 5): Generate full description
6. Copy.AI Request (Module 6): Generate metadata (tags, hashtags, excerpt)
7. Text Parser (Module 7): Extract tags and hashtags
8. JSON Compiler (Module 8): Assemble all outputs
9. Google Sheets or HTTP Post (Module 9): Save results

The Manual Alternative

If you want more control over the output at each stage before it moves forward, you can introduce approval gates.

Add a "Pause" module after Step 2 (the summary). This lets you review Resoomer's output and manually edit it if needed before it feeds into title and description generation. Simply click "Resume" in Make's execution history when you're ready.

Similarly, you might pause after Copy.AI generates title options, pick your preferred three, and only feed those into the next step. This gives you creative oversight without automating away quality control.

The cost is a few minutes of human time per video, but you're no longer doing the transcription-to-summary-to-description grind from scratch.

Pro Tips

1. Rate Limiting and API Keys

Each API has rate limits. Copy.AI allows 100 requests per day on the free tier, while Quick Creator and Resoomer charge per request. To avoid hitting limits mid-batch, add a 5-second delay between API calls in Make using the "Sleep" module. If you're processing 10 videos weekly, you'll stay well within free tiers; if you're doing 50+, consider upgrading.

2. Handle Transcription Errors Gracefully

If a transcript is corrupted or incomplete, Resoomer and the other tools will struggle. Add error handling: if Resoomer's API returns an error or a summary under 50 words, trigger a notification email to you rather than proceeding. Use Make's "Router" module to branch the workflow on error.

3. Store Everything for Reuse

Don't discard intermediate outputs. Save the summary, title options, and description to a Notion database or Airtable base alongside the original transcript. Future videos on similar topics can reference this repository, and you'll spot patterns in what titles perform well.

4. Customise Temperature Settings for Different Content Types

Copy.AI's temperature parameter controls creativity. Set it to 0.5 for technical or educational content (more consistent, less creative), and 0.8 for entertainment or lifestyle content (more varied, punchier). Quick Creator similarly benefits from tuning based on your channel's tone.

5. Monitor Costs Weekly

Each API call costs pennies, but they accumulate. Quick Creator is typically the most expensive per request. Set up a simple spreadsheet tracking costs: (API calls × cost per call) × videos per month. At 10 videos weekly, you're looking at roughly £15-£30 monthly if you're on paid plans. Free tiers usually cap out around £0 until you hit volume, so start there and upgrade only when you need to.

Cost Breakdown

This assumes you're providing transcripts yourself. If you need transcription (via AssemblyAI, Rev, or similar), add £10–£50 monthly depending on video length and volume.