Found something good?

A clipping page that stopped uploading still pulled 173,123 views in the last 7 days.

YouTube says the channel is worth $32 a month.

Both of those numbers are lying to you. And as of this week, the machine that exploits the gap between them runs on a model Anthropic released 48 hours ago.

Let me show you the whole thing: the case study, the math, and the exact setup.

The $32 channel that won't die

There is a YouTube channel called Iman Gadzhi Clips. 80.7K subscribers. 1.9K videos. Upload frequency right now: roughly zero per week.

The page is basically dormant. And the vidIQ panel still shows 173,123 views gained in the last 7 days, up 18.89%.

Estimated monthly earnings: $32.

Scroll the catalog and you see why it refuses to die. Single clips sitting at 5.4M views. 4.6M. 3.8M. 2.9M. 2.2M. A library of moments that keeps getting served years after it was cut.

If you think the business is YouTube AdSense, this channel is a joke. 173K weekly views for $32 a month is not a business. It is a vending machine that pays out in coins.

But $32 is just the rent YouTube pays for attention it resells to advertisers. The real price of attention gets set somewhere else, and the gap between those two prices is the entire clipping economy.

Nobody understands that gap better than the man being clipped on this channel. He did not stumble into it. He priced it, three different times.

The guy who industrialized clipping

Most people know Iman Gadzhi as the agency guy. That is the wrong frame. Gadzhi is the guy who monetized clipping three separate times, on three different sides of the same trade.

First, as the creator being clipped. In August 2022 he ran what he called a social media takeover: 100,000,000 views in a single month and roughly 1 million new followers in 30 days. He had around 600K YouTube subscribers when he did it. He sits at 5.96M today.

That growth was not one viral video. It was armies of clippers flooding Shorts, Reels, and TikTok with his moments, all at once.

Second, as the buyer. Clipping campaigns tied to Gadzhi run on Whop's Content Rewards, where anyone can cut his content and get paid per 1,000 views.

Typical campaign rates sit around $1 to $3 per 1,000. You find the moment, you post it, you submit the link, you get paid.

Third, and this is the part almost nobody connects: on April 2, 2025, Gadzhi became co-owner and investor of Whop itself. The marketplace that processes the clipper payouts. Alongside Peter Thiel, The Chainsmokers, and Insight Partners.

Whop processes over $1B in payments annually, and its average creator earns $8,413 a month.

Read that arc again. He got clipped. Then he paid for clipping. Then he bought the toll booth.

Meanwhile the ecosystem around his catalog kept compounding without him touching it. Fan clip channels built on his content run from the dormant 80K page above to monsters at 2.5M and 11.4M subscribers. An entire economy mining one man's backlog.

That is what a clipping flywheel looks like when it is run with intent. In 2022 it took an army of editors. Today it takes one operator and one model.

The $24,818 math

The number in the title is not magic. It is the same distribution math from my last guide, scaled to what an agent-run page can actually push.

If a clipping operation gets toward 6,000,000 tracked views a month, and the backend pays around $4.14 per 1,000 views, that lands near $24,818 a month. The exact cents do not matter. The spread does.

That backend payout can come from a creator campaign, a Whop content reward, an affiliate deal, a paid community, or a client who pays for distribution because they care about booked calls, not AdSense.

Without the backend, those exact 6M views are the $32 channel. The views never change price. The destination does.

6M views a month sounds heavy until you break it down: roughly 200K views a day across YouTube Shorts, TikTok, Instagram, and X combined. Pages mining a strong catalog do this with 6 to 8 clips a day. The bottleneck was never the math. It was the labor.

That is the part that died this week.

What changed this week

Anthropic released Claude Fable 5 on June 9. It is the first Mythos-class model the public can touch, and Anthropic says its capabilities exceed every model they have previously made generally available.

The headline benchmarks are not the point for us. Two things are.

One: it is built for operator work. Long tasks, tool use, judgment calls, staying on plan without a human babysitting every step. It runs 25 to 30% faster than Opus 4.8 and finishes work in fewer turns.

And running a clipping page is operator work: which video is next, which clips deserve to exist, which platform gets what, what is scheduled, what won. The cutting was automated months ago. The running of the whole loop is what just became automatable.

Two: the window. Fable 5 is included in Pro, Max, and Team plans at no extra cost until June 22. After that it moves to usage credits.

On the API it costs 2x Opus 4.8. So you have 11 days to run the most capable publicly available model on your clipping workflow for free.

In 2022, Gadzhi's takeover needed dozens of humans cutting and posting in parallel. The 2026 version of that machine is three parts, and you can set it up in about ten minutes.

The stack

Part one: Claude Fable 5 is the operator. It runs the loop, makes the judgment calls, and uses the tools.

Part two: Vugola is the clipper. You hand it a long-form video URL, it finds the strong moments, cuts them into 9:16 clips, burns captions in your chosen style, and scores every clip for virality from 0 to 1. Jobs take 10 to 30 minutes and the agent can poll, download, and rank everything without you watching a single timeline.

Part three: Postiz is the distribution engine. It schedules posts across 28+ platforms, and it ships something that matters a lot more in the Fable 5 era: an agent Skill.

My previous guides wired everything through MCP servers and config files. That still works. But the new pattern is Skills, and the difference is worth ten seconds of your attention:

MCP = gives the agent ACCESS to a tool
Skills = teach the agent the WORKFLOW for the tool

An MCP server hands Claude a set of tools and hopes it figures out the order. A Skill is a file that teaches it the whole playbook: which command first, what the output means, what to do when something fails. One command to install. No JSON. No config file.

Fable 5 reads both. Here is the full setup.

Set it up (about 10 minutes)

Do this once. Keep real keys private, use placeholders in anything public, and never paste secrets into screenshots.

Step 1: Run Fable 5.

In Claude Code, switch models:

Or type /model inside a session and pick Fable 5. On Claude Desktop, select Fable 5 from the model picker. Until June 22 this costs you nothing extra on a paid plan.

Step 2: Connect Vugola (the clipper).

You need Node.js 20+ and a paid Vugola account. Generate a key at vugolaai.com/dashboard/api-key. Keys start with vug_sk_. Sanity-check it:

It returns your plan and remaining credits. One credit equals one minute of source video.

For Claude Code:

For Claude Desktop, the one-command installer edits your config for you:

Always pin the version. There is also a downloadable Claude Code Skill on the Vugola docs page if you want the workflow taught the Skills way too.

This gives Fable 5 the clipping tools: clip_video, get_clip_status, download_clip, caption_video, get_usage.

Step 3: Connect Postiz (the distribution).

Three commands. This is the whole install:

The login is a device flow. A one-time code shows in your terminal, your browser opens, you authorize, done. No client ID. No secret. No config file.

Verify everything is wired:

If your connected social accounts come back, the machine is assembled. Fable 5 now has eyes on your Vugola credits and hands on your posting calendar.

Hand it the keys

The prompt that runs the page is not a creative brief. It is a work order.

You read the calendar, veto what deserves a veto, and go do something else. The page posts whether you are at your desk or not.

What Fable 5 should judge

Here is where most people will waste this entire setup: they will clip everything and post everything. A machine that posts garbage faster is not an edge. It is a faster way to teach the algorithm your page is garbage.

Look at what actually wins on the Gadzhi clip pages. It is not the clean tutorials. The top of the catalog is Iman Gadzhi's BIGGEST Lie at 2.9M. Buying a Bugatti at 412K. I have a Maybach. Why he keeps buying watches. Taxes. Relationships. Red flags.

Status, money, conflict, confession, disbelief. The moments a stranger cannot scroll past.

Every niche has its own version of this. Fitness pages clip shame and transformation. Trading pages clip risk and flexes. Podcast pages clip confession and conflict. The agent removes the labor. Your taste decides what deserves to exist.

Vugola's virality score does the first pass for you. Your job is the veto.

The honest version

I am not going to pretend you flip this on and $24,818 appears in month one. Here is the conservative math, because the inflated version helps nobody.

A new page in its first 60 to 90 days realistically does 300K to 1.5M tracked views a month, not 6M. If your only backend is public Whop campaigns at $1 to $3 per 1,000, that is $300 to $4,500 a month. The $24,818 number needs two things the dormant Gadzhi page proves are possible but not automatic: a deep catalog that compounds, and a backend that pays above campaign rates, meaning your own offer, an affiliate deal, or a client paying for distribution directly.

So why bother? Because the cost side just collapsed. The 2022 version of 6M monthly views was a payroll. The 2026 version is a Vugola subscription, a Postiz subscription, and a model that is free until June 22. Even the conservative $1,500 a month is a real margin when the labor is one prompt a day.

And one more honest line: do this with permission, a campaign, or a deal. Clipping creators who never agreed to it is building a business on a cease and desist.

Your first 72 hours

Day 1: do the 10-minute setup above. Pick one creator with a deep backlog and one backend. Run your first long video through Vugola and just read the virality scores against your own gut. Calibrate.

Day 2: let Fable 5 schedule the top 6 clips through Postiz, 2 a day. Veto anything weak. Your veto is the only part of this machine that cannot be installed.

Day 3: pull the first signal and make the loop close itself:

postiz analytics:platform <integration-id> -d 30

Ask Fable 5 to tell you what won by hook, topic, and platform, then feed that straight into the next clipping job. From here the machine improves every time it runs. Most people quit clipping in the gap between posting and learning. The gap is gone.

TL;DR

Get the tools

Everything in this guide is live today. Two links and you have the whole machine:

Vugola turns a long video into ranked, captioned, ready-to-post clips. The clipping engine in this entire workflow. Start here:

https://www.vugolaai.com

Postiz is how your agent posts to 28+ platforms with a single skill install. The distribution layer. Get it here:

https://postiz.com

Run them on Fable 5 while it is free. After June 22 you will wish you had these 11 days back.

thanks for reading