Opens on the claim instead of clearing its throat first. Short paragraphs, no block of hashtags, no “Thoughts?” at the end.
FOR YOUTUBE CREATORS
You spent three weeks on a
video nobody watched.
It wasn’t the thumbnail, and it wasn’t the edit. That video was decided long before you hit record — and you had no way of knowing.
No card. One complete run, on us.
- Videos scored
- 975
- Outliers found
- 107
- Audience analysed
- 121.5Msubscribers
- Biggest outlier found
- 63.7×DiscoveryChannelInd
HOW IT USUALLY GOES
So you do what everyone does.
You go and look at what’s working.
You open the channel that’s beating you in your niche. You scroll the uploads, sort by most popular, and write down the titles that pulled millions. Somewhere in there, you tell yourself, is the pattern.
It isn’t. What you found is their biggest moments, which is a different thing — and mostly it tells you they have more subscribers than you do.
Because a view count means nothing on its own. Here are two channels Cutline has actually analysed.
Is a 5.5M view video good?
Veritasium
Typical video 10.1M
0.5×
Their worst upload in months.
fernTypical video 3M
1.9×
The biggest thing they ever made.
Same number. Opposite lessons. Sorting by views cannot tell them apart — and from the outside, neither can you.
THE PART NOBODY TELLS YOU
Every channel is already
telling you what works.
Just not in its top ten. It’s in the gap between that channel’s best videos and its ordinary ones — the only honest signal on YouTube, because it cancels out the one variable that wrecks every comparison you’ve ever made: audience size.
A video that beat its own channel four times over did something the other uploads didn’t do. Find enough of those on one channel and the pattern stops being your opinion and starts being evidence.
That is the entire idea behind Cutline. It reads a channel against itself.
SEE IT FOR YOURSELF
Here is what that looks like
on channels you know.
Real reports, cached exactly as Cutline produced them. Open any one — nothing on this page is a mockup.
CHANNEL REPORTVox
- Subscribers
- 12.7M
- Typical video
- 46.2K views
- Best video did
- 19.2× that
CHANNEL REPORTVeritasium
- Subscribers
- 21M
- Typical video
- 10.1M views
- Best video did
- 3.5× that
CHANNEL REPORTfern
- Subscribers
- 5.2M
- Typical video
- 3M views
- Best video did
- 4.5× that
CHANNEL REPORTJohnny Harris
- Subscribers
- 7.9M
- Typical video
- 3.3M views
- Best video did
- 2.2× that
CHANNEL REPORTKurzgesagt – In a Nutshell
- Subscribers
- 25.4M
- Typical video
- 6M views
- Best video did
- 2.6× that
CHANNEL REPORTMagnatesMedia
- Subscribers
- 1.9M
- Typical video
- 1.2M views
- Best video did
- 10.2× that
WHAT HAPPENS WHEN YOU PASTE A LINK
Four minutes later,
you’re writing.
- 01It reads the channel against itself
Every long-form upload it can see, scored against what a normal video does on that channel. Shorts are thrown out — they have their own physics and mixing them in ruins the median.
- 02It works out what the winners share
Not a list of titles. A written read of the angle, the promise and the shape the breakouts have in common, and what a different creator should take from it.
- 03It writes six ideas and scores them
Original ideas built from the pattern, never rewrites of their videos. Each one carries a number and the reasoning behind it.
- 04It writes the script
Three to twenty minutes, hook and structure included, in your voice. Don't like one paragraph? Leave a note on that paragraph and only it gets rewritten.
- 05It turns that into a week of posts
The finished script becomes posts for LinkedIn, X, Instagram and TikTok — written for each one rather than pasted between them, and TikTok gets a spoken hook and on-screen text rather than a caption.
A whole run costs 73 credits. You start with 100.
THE OBVIOUS OBJECTION
Six ideas is easy.
Six good ones isn’t.
Any model will hand you six ideas. Most will be one idea wearing different hats, and you get no way to tell which of them is worth three weeks of your life.
So Cutline scores them — out of 100, on how likely each is to beat that channel’s median, with the reasoning that produced the number. And it is not allowed to flatter you: at most two ideas may score above 80, and at least one has to come in under 60.
- 85/100
Why the world’s most secure vault is built inside an active volcano - 82/100
How a single spreadsheet error bankrupted a municipal government in 2024 - 55/100
Why urban planners in Tokyo refuse to install public trash cans
Real ideas Cutline generated for Vox.
AND THE OTHER OBJECTION
Then it comes back sounding
like everybody else.
That is the real reason people abandon AI writing tools. The script is competent, agreeable and completely anonymous, and you can hear it in the first sentence. Nobody wants to put their face on that.
So Cutline asks six questions before it writes a word for you.
- Who are you writing for?
- What do you help them achieve?
- How should it sound?
- What should the content drive toward?
- What does your field get wrong?Most consultants compete on price because they never learned to sell the outcome…
- Anything you never want said?synergy, hustle, our proprietary framework
The last two carry it. An opinion is what makes writing sound like a person, and a list of words you refuse to say is what stops it sounding like everyone else in your niche.
AND IT DOESN’T STOP AT THE VIDEO
One script.
Twenty posts.
The script is the expensive part, and most creators spend it once. The same argument that carried a ten-minute video is still good on a Tuesday afternoon in a feed — it just has to arrive in the shape that feed expects.
So the last step takes the finished script to four platforms, up to five posts each. Not the same paragraph relabelled four times: each one is written for where it lands, and a post that would read identically on two of them is treated as a failure.
X
Numbered threadA first line that has to survive alone in a timeline, and every post after it earning the one that follows.
Written slide by slide at six to eight words each, because a paragraph cut into squares is not a carousel. Caption written last, and separately.
TikTok
Short-form videoNot a caption. The spoken hook for the first three seconds, thirty to forty-five seconds written to be said rather than read, and the on-screen text to run alongside it.
Already have the script? You can start here instead — paste it in and skip everything above.
AND IF YOU ALREADY KNOW WHAT YOU’RE MAKING
You don’t have to start
from a channel.
The whole run
Outliers, the pattern, six scored ideas, a script, and the posts.
A finished script
Skip the research. Describe the video and it writes it in your voice.
A week of posts
Paste something you already wrote and take it to four platforms.
The next three weeks
should go somewhere.
One hundred credits, free, no card. That is a full run: a channel read against itself, six ideas scored, a script written in your voice, and a week of posts out the other end. Enough to find out whether the next one lands.