There is a way to reuse a LinkedIn carousel you admire without stealing a word of it. AI Central's method has two halves. Download the carousel as a PDF, hand it to ChatGPT with a ghostwriter prompt so the model reports back the structure that makes it work, then drag the same PDF into Canva, which rebuilds it as an editable template with the fonts intact. You supply the new topic. The skeleton does the rest.
Reviewed August 2026.
The part of a carousel worth copying is the skeleton
AI Central files this under a deliberately uncomfortable label, the forbidden technique, and the discomfort is the point. Most people trying to learn from a carousel that performed well end up imitating its sentences, the one part that does not transfer. What transfers is the architecture underneath, the order of the slides, the density of each one, the typeface, the rhythm of hook and payoff.
That is why AI Central's very first instruction is not to hunt for the highest performing carousel on the platform. It is a fit test.
Choose one close to your brand style
Read that as the boundary of the whole method. You are borrowing a chassis, not a body of work, and if the source already looks like something your audience expects from you, the structure you inherit will not fight your voice when the new copy lands in it.
The first prompt asks for a read, not a rewrite
This is the step people skip, and skipping it is what turns the technique into plagiarism. AI Central has you upload the source PDF and ask ChatGPT to analyze it before asking it to produce anything. The role assignment comes first.
You are an expert ghostwriter that specializes in LinkedIn marketing and you have years and years of experience on this platform.
Then a single instruction, please analyze this LinkedIn carousel. The uploaded file is the entire payload. No topic yet, no brief, nothing to write. The ghostwriter framing matters because a ghostwriter's job is to read someone else's format and reproduce its effect, not its content, which is exactly the output you want here.
AI Central sets the expectation for what comes back.
You’ll receive a concise, point-by-point breakdown.
What the breakdown gives you
AI Central runs the technique on a real example, a twenty-one page carousel titled 7 Easy Steps To Build Your 1st GPT Fast, credited to Artur Sossin, an AI growth strategist, and Sabahudin Murtic, who works in AI content creation. The model returns its report under a heading it writes itself, Carousel Breakdown and Performance Analysis, and sorts the findings into what works well.
Four things make that list. A strong hook on the opening slide. Clear problem framing across slides two and three. A step-by-step format running from slide four to slide ten. And minimalist visual design, which the model summarizes as a clean layout with plenty of white space and hierarchical formatting that makes the content skimmable.
The useful part is that the model explains why each one works rather than just naming it. On the cover, it isolates the audience targeting in a single line of subhead copy.
Subhead “More results in less time for PMs” targets a defined audience (Product Managers), which is key to LinkedIn resonance.
On the problem slides, it does the same for the pain points the original chose to name.
Identifies relatable pain points: “Responses that look misaligned,” “Settings you’re unsure how to tweak.”
AI Central's note underneath that output is the quiet payoff of the whole first half.
As a bonus, you’ll have the structure.
The second prompt, read constraint by constraint
With the analysis sitting in the thread, AI Central's second prompt does the writing. Every clause in it is load-bearing, so it repays reading slowly.
It asks for short, crisp sentences and readability enhanced by a captivating format. It tells the model to take its time for thoughtful composition, a pacing instruction more than a quality one. It bans hashtags outright, caps the opening statement at eight words, and asks for each part to be explained step by step. Then it points back at the file already in the thread.
Use the same tone and structure as in the carousel I’ve provided.
That sentence is the hinge between the two prompts. Without the analysis step, the model has nothing concrete to imitate and defaults to generic LinkedIn cadence. With it, tone and structure have a documented referent.
The eight word cap earns its own attention as the only constraint in AI Central's prompt carrying a hard number. A carousel cover is read at thumb speed, and giving the model a number rather than an adjective is what stops it returning a fourteen word headline that it calls punchy.
In the worked example the output cover reads 7 Ways To Win On LinkedIn Fast, over a three beat line, real growth, real results, real strategy. Slide two states the problem, slide three turns toward the fix, slide four opens step one, optimize your profile, with a professional photo and a clear, benefit-driven headline. The shape of the source is fully intact. None of its sentences are.
AI Central also covers the case where you have a format but no content. Ask ChatGPT for the raw list first, in the example seven steps to growth on personal branding, then feed that into the carousel prompt. The guidance attached is one line.
You can use it with your niche also
The Canva half is where the design actually crosses over
This is the half most people never reach, and the reason AI Central pairs the two tools rather than treating this as a prompting trick. Open Canva, start from a free template, then drag the source PDF into the editor. Canva offers to apply the imported design across the whole file, all twenty-one pages in AI Central's example.
What survives the import is worth stating precisely, because it sets your production budget for the rest of the job.
- The text becomes directly editable inside Canva, so you replace copy rather than rebuild slides.
- The fonts used in the original come across with the layout.
- Images do not transfer. AI Central flags that limitation explicitly.
That last constraint is the honest cost of the method. Layout, typography and slide rhythm import cleanly. Artwork does not, so anything illustrative has to be replaced with your own. Which is, conveniently, the part a rights holder would care about most.
Two things AI Central will not let you drop
Whatever else changes, two structural elements stay fixed. A clear hook at the beginning, and a call to action at the end.
Both survive for the same reason. The first slide decides whether anyone swipes at all, the last decides whether the swipe becomes a follow. In the example, the closing slide is a follow prompt naming both creators and the subject they cover, which is the entire return on the twenty slides before it. Strip either one and you have a document, not a carousel.
What to do with this
- Pick your source for fit, not for reach. AI Central's first instruction is to choose one close to your brand style, and that is also the version you can defend if the original creator sees your post.
- Always run the analysis prompt before the writing prompt. The model cannot match a tone and structure it was never asked to examine.
- Hold the eight word rule even when the model pushes back. Regenerate rather than negotiate.
- Budget time for artwork. Text and fonts import into Canva, images do not.
- Change the words, keep the scaffolding. If a sentence from the source survives into your version, you have crossed from technique into copying.
Frequently asked questions
Is this just stealing someone else's carousel?
Not if you run it as AI Central describes it. The copy is generated fresh from your own topic, the images have to be rebuilt because they do not import, and what you reuse is structure, typography and slide order. Keep the source's sentences out of your version and you are learning from a format, which is what every designer has always done.
Which ChatGPT model do I need for this?
AI Central's walkthrough runs the carousel through ChatGPT-4. The practical requirement it implies is a model you can upload a PDF to, since the file itself is the input for the analysis step. If the model cannot read the file, there is nothing for the second prompt to imitate.
Why does the prompt limit the opening line to eight words?
Because it is the only instruction in the prompt with a hard number, and hard numbers are what models obey. Eight words is tight enough to force a real hook instead of a description of one.
Do the images come across when I import the PDF into Canva?
No. AI Central is explicit that images do not carry over. You get directly editable text and the fonts used in the original, enough to inherit the layout but not enough to skip the design work on any illustration or screenshot.
What if I have the format but no topic to put in it?
Ask ChatGPT for the content list first. AI Central's example is a request for seven steps to growth on personal branding, which returns a numbered list you can drop straight into the carousel prompt. Swap the subject for your own niche and the rest of the process is unchanged.