👋 Hey there, I’m Richard. Each week, I tackle reader questions about creating world-class events, brand building, and accelerating your career.
Today, I want to show you one of the smartest, most practical uses of AI in event marketing right now: a content repurposing agent system that automates blog posts, social content, and more - using your long-form videos as the engine.
I built this to save time, but it ended up saving thousands in freelancer and ops costs too. And the best part? It doesn’t just spam out low-quality AI content. This system blends automation with human oversight, making it ideal for marketers who care about quality, brand voice, and trust.
In this post, you’ll learn:
What AI agents actually are (and how they differ from prompts)
Why multi-agent systems matter for marketers and creators
A breakdown of the repurposing system I built
How to build and run your own - without writing a line of code
Let’s get into it.
First, What Is an AI Agent?
If ChatGPT is a helpful assistant waiting for prompts, an AI agent is a proactive operator. It can:
Trigger tasks from new data (like detecting a new YouTube upload)
Use tools autonomously (transcription, publishing, formatting)
Make decisions and delegate tasks
Loop humans in for review and approvals
This means agents don’t just answer. They execute - across an entire workflow.
If you’ve ever wished for a mini marketing ops team to help you get more from your content, multi-agent systems make that possible.
Why AI Agents Matter for Marketers
Marketers today are stuck between two extremes:
Doing everything manually - writing, formatting, posting, promoting
Over-automating with low-quality AI content
Agents offer a better path.
You still create the source material - a podcast, a blog, a webinar - but instead of slicing it manually, your agents:
Transcribe it
Repurpose it for different platforms
Format and post it
Send everything back to you for final sign-off
This means you keep creative control while massively reducing the execution load. And because it’s built on your own content - not ChatGPT hallucinations - it reflects your voice, expertise, and goals.
Inside My Repurposing Agent System
I built this system using Relevance AI - a no-code platform for building AI agents - and Make.com to handle integrations and publishing.
Here’s how it works.
The Manager Agent
At the core is the YouTube Repurposing Manager Agent. It’s the orchestrator. When I upload a new video, it:
Triggers the workflow
Sends the link to specialist sub-agents
Collects and routes feedback
Publishes final versions via Webflow or LinkedIn (after I approve them)
The Sub-Agent Team
Each sub-agent handles one specific task:
Blog Writer Agent - transcribes the video and writes a high-quality blog post
LinkedIn Writer Agent - creates native social content
Email Writer Agent (optional) - drafts newsletter copy
Evaluation Agent (future) - reviews outputs for quality and consistency
Each sub-agent is set to auto-run, meaning once the manager delegates a task, they handle it end-to-end and return the result.
Human-in-the-Loop Review
This is critical. I’ve built a review loop directly into the manager agent. Once a sub-agent finishes its draft:
It sends it back to me for review
I can suggest edits (e.g., tone, clarity, structure)
The manager sends that feedback back to the sub-agent
The agent revises the content and resubmits
Only once I give final approval does the system publish anything.
What This System Actually Does
Let’s walk through the full workflow for a single YouTube video:
Trigger - A new YouTube video is detected via Make.com
Transcribe & Draft Blog - Blog writer transcribes the video and writes the post
Review - Manager agent sends it to me for feedback
Revise - Blog writer incorporates feedback
Format - The post is converted to HTML with proper styling
Publish Blog - Webflow tool posts the final article to my website
Draft Social - LinkedIn writer crafts content from the same video
Review & Approve - I make small edits, upload visuals, and hit publish
In a few hours, one video becomes a blog, a LinkedIn post, and optionally an email - with minimal human friction.
Why This Matters for You
If you’re a:
Social manager for a event who must 10x content without 10x effort.
A B2B event marketer sitting on hours of session footage.
Event organiser with a library of keynotes/breakout sessions.
This kind of system gives you leverage. It turns one core piece of content into many - consistently, professionally, and with human oversight.
Even better, it saves you:
Countless hours of admin and formatting
Thousands of dollars in freelancer costs
Mental overhead from switching between platforms and tools
And because you’re still involved in reviewing content before it goes live, you don’t compromise on brand, tone, or polish.
How to Build Your Own (No Code Needed)
You’ll need just three tools:
Relevance AI (to build agents)
Make.com (to connect systems and post content)
Your content (YouTube, podcasts, webinars, etc.)
Here’s the high-level setup:
Build a Manager Agent — define its SOP (standard operating procedure), delegate tasks, and add approval steps
Create Specialist Sub-Agents — one for blogs, one for LinkedIn, one for email; each with a specific prompt and tool
Use Make.com Webhooks — to automate publishing to Webflow or LinkedIn
Trigger Automatically — when new content is uploaded, it kicks off the flow
No code. No API headaches. Just clear logic and modular tasks.
Watch the Demo
Want to see exactly how this system runs in real time?
🎥 Watch the full demo below where I walk through:
The agent dashboard
A real YouTube repurposing workflow
How I built each agent
The Webflow and LinkedIn publishing integrations
Final Thoughts
AI agents are the next big unlock for content teams.
They’re not here to replace your creativity — they’re here to amplify your execution. To turn a single video into a multi-channel content strategy. To remove the bottlenecks between ideation and publishing.
If you’re already creating valuable content, this system turns it into an engine.
I hope this breakdown helps you see what’s possible. Now go build your own — and spend less time formatting blog posts.
Until next time,Richard
