Why I Built an AI Video Content Engine

I love creating things.
Videos. Talks. Articles. Workshops.
But you know what I don’t love?

Spending hours tweaking subtitles, writing captions, drafting LinkedIn posts, and wondering if I already summarized that video better two months ago.

At some point, the admin work of creativity starts eating the joy of creativity.

That’s when I realized: I needed a second brain.

Not Just Automation - Amplification

I didn’t want to automate creativity away.
I wanted to amplify it.

That’s when I started experimenting:

What if I could upload a transcription and get back not just text, but a blog post, a quiz, and a LinkedIn post?
What if I could run a video through a workflow and have the first draft of everything waiting for me?
What if AI could help me start faster, so I could focus on polishing, connecting, creating?

The answer?
A modular, scalable, AI-powered content engine.

How It Works (In a Nutshell)

I designed a system that looks like this:

Upload a video (or use a recording)
Trigger a Step Functions workflow
Auto-transcribe the audio
Auto-detect key scenes with Rekognition
Invoke Bedrock (Claude model) to generate:
- A blog post
- A LinkedIn-style post
- A short quiz
- A concise summary
Save everything into S3 for easy reuse

Simple on paper.
Magical when it works.

Workflow Overview

Why It Matters

Building this wasn’t about automating my job away.
It was about giving myself breathing room to stay creative longer.

Instead of staring at a blank screen, I now get to iterate.
Instead of worrying about transcription errors, I get to refine stories.

I started this project because I needed it.
But the more I used it, the more I realized:

Every creator deserves a crew.

Even if that crew is made of pixels and prompts.

In the next post, I’ll show you exactly how I designed the architecture - and why choosing modular pieces (like AWS Transcribe, Bedrock, and Rekognition) made all the difference.

Thanks for being here

Questions? Ideas? Let’s talk: hello@lindamohamed.com