@olivermerrick___'s reel — transcript & breakdown

Plays

122.9K

Likes

3.7K

Comments

509

Engagement Rate

3.39%

Create your own version

Use this post as inspiration to create your own

Remix
AI can already cut, caption and format a clip for you. Free GitHub skills teach Claude the editing part in minutes.

So why isn’t everyone going viral off podcast clips?

Because editing was never the hard part. Choosing the moment is. AI does not know which 47 seconds out of a two hour conversation deserve to become the clip. No context. No content intelligence.

Here is how you build that layer.

Scrape the latest 100 clips from the biggest podcast short form accounts. Steven Bartlett, Chris Williamson, Alex Hormozi, anyone running serious long form volume. Capture views, post date, duration, transcript, text overlay and format.

Then scrape their original long form episodes and pull the complete transcripts with timestamps. The timestamps are non negotiable. They are what lets Claude match every clip back to the exact episode and the exact moment it came from.

Now you reverse engineer the decision.

Why those 47 seconds. Where the clip starts. What context got removed. Whether the payoff moved forward. What spoken hook, text overlay and format they used. And how it performed against that account’s normal clips.

Run enough of these and the patterns surface. Turn the patterns into a Claude skill that understands how the best podcast teams select and package their clips.

Then hand it any new timestamped transcript. It returns the strongest moments, exact timestamps, hook options, text overlays, formats and editing instructions. The editing skill does the easy part.

None of this needs new software. The tools exist. The clipping skills exist. The data is public. Nobody is assembling it into an intelligence layer.

Impossible to fail does not mean every clip goes viral. It means every clip starts with real evidence, and every result makes the next selection smarter.

Build the selection layer first. The edit is the easy bit.
1:27

Audio

Calm lo-fi beat playing in the background

Lo-fiChillSlow

Caption

AI can already cut, caption and format a clip for you. Free GitHub skills teach Claude the editing part in minutes. So why isn’t everyone going viral off podcast clips? Because editing was never the hard part. Choosing the moment is. AI does not know which 47 seconds out of a two hour conversation deserve to become the clip. No context. No content intelligence. Here is how you build that layer. Scrape the latest 100 clips from the biggest podcast short form accounts. Steven Bartlett, Chris Williamson, Alex Hormozi, anyone running serious long form volume. Capture views, post date, duration, transcript, text overlay and format. Then scrape their original long form episodes and pull the complete transcripts with timestamps. The timestamps are non negotiable. They are what lets Claude match every clip back to the exact episode and the exact moment it came from. Now you reverse engineer the decision. Why those 47 seconds. Where the clip starts. What context got removed. Whether the payoff moved forward. What spoken hook, text overlay and format they used. And how it performed against that account’s normal clips. Run enough of these and the patterns surface. Turn the patterns into a Claude skill that understands how the best podcast teams select and package their clips. Then hand it any new timestamped transcript. It returns the strongest moments, exact timestamps, hook options, text overlays, formats and editing instructions. The editing skill does the easy part. None of this needs new software. The tools exist. The clipping skills exist. The data is public. Nobody is assembling it into an intelligence layer. Impossible to fail does not mean every clip goes viral. It means every clip starts with real evidence, and every result makes the next selection smarter. Build the selection layer first. The edit is the easy bit.

Hook

AI CLIPPING SYSTEM *full setup tutorial*

0:00Hook

The video starts with a bold claim about an 'AI Clipping System' that is 'scientifically impossible to fail,' immediately grabbing attention and establishing a high-value proposition.

“Here's how to build an AI clipping system so good it is scientifically impossible to fail.”

0:05Social Proof

The creator leverages the success of well-known figures like Steven Bartlett and Alex Hormozi, suggesting they 'already use something like this,' which builds credibility and implies effectiveness.

“I would bet teams behind Steven Bartlett and Alex Hormozi already use something like this.”

0:12Setup

The video introduces the problem that AI can cut videos but lacks 'context' and 'content intelligence' to identify viral moments, setting up the need for the 'Clip Brain' solution.

“There are so many free skills like Remotion that let Claude code edit videos programmatically like add captions and remove the blanks. So if clipping is so easy, why isn't everyone going viral that does it? Because the AI knows how to cut a video. But it doesn't know the exact moment that needs to be a clip. It has no context and no content intelligence.”

0:31Content

The core 'build' process is explained step-by-step, involving scraping short-form clips, long-form transcripts, matching them, and analyzing patterns using an 'AI Clip Brain' to understand viral hooks.

“Alright, let me show you how to build this thing. First, you're going to scrape any short-form account and get all of their clips. Just use any scraper like Apify and make sure to pull everything like views, likes, comments, etc. Then you also scrape the long-form with the transcript and the timestamps. And just ask Claude to match the short-form transcripts to the long-form video like find the source. Now after you've done that whole setup, you can ask it to start matching what moments and sources and what the patterns are between all the clips. So you can use the Analyze skill to literally see which clips do better and why. This hook does so much better than this hook. Like here's the proof again, it happens so consistently.”

1:09Recap

The creator quickly summarizes the entire 'tech stack' required for the system, making it seem accessible and actionable while reiterating its potential.

“You can build the ultimate clip brain internally for your business. Like all you need is this tech stack: Apify to scrape, Claude to run everything, store it all in Superbase, find the skills on GitHub, and then you can have your actual content ideas on Notion and Remotion to do all the editing.”

1:26CTA

A direct question is used as a call to action, engaging the audience and prompting interaction for further content, ensuring continued viewership.

“Do you want me to drop the full system?”

Details

Account@olivermerrick___
Posted (UTC)
Date (UTC)Jul 28, 2026
Duration87.4s
Last synced (UTC)

More viral reels, analyzed