One long episode is roughly ninety minutes of raw material, and most repurposing attempts waste it. The usual first instruction is "turn this video into twelve posts", which produces twelve versions of one document, each too thin for the platform it is aimed at.
A content repurposing ai prompt works when it runs in two stages: extract the discrete claims from the source, then assign each claim to the format that can carry it. Below is the extraction prompt, the twelve-asset map, and the checks that catch the two defects that make repurposed content perform badly, which are duplicated ideas and posts that ignore their platform.
Stage one: extract the atoms before you write anything
Atoms are the standalone claims, examples and opinions in your source. Each one is small enough to be true of itself and interesting enough to carry a short post on its own.
Read the source material below and extract the atoms it contains. An atom is a single claim, example, story, opinion or step that stands on its own without the rest of the episode. For each atom give: the atom in one sentence, the timestamp or section it came from, what the listener would need to know for it to make sense, and whether it is a claim needing verification, an opinion, or a story. Aim for [number] atoms. Do not invent anything that is not in the source. Do not merge two atoms into one post. If the source is too thin to produce that many distinct atoms, say so and give the number you can actually support.
Read this step carefully, because the model's output is the entire raw material for everything that follows. Requesting more atoms than the episode contains is how twelve posts become five posts with padding.
Stage two: map the atoms onto twelve formats
Each format has a shape and a length. Assign the atom that fits rather than asking the model to fit an atom into a format.
Using the atoms I extracted, write twelve assets. Each asset must use a different atom and a different format. For each give: the atom used, the format, the platform it is for, a draft, and the one thing that makes it worth that platform's reader's time. The twelve assets: (1) a short LinkedIn post of 100 to 150 words, (2) an Instagram carousel of eight slides with a caption under 150 words, (3) a thread of eight posts, (4) an email to your list under 200 words, (5) a short blog section of 400 words, (6) three standalone quotes with context, (7) five questions your audience has been asking, (8) a short video script under 90 seconds built from one atom, (9) a longer video script of 5 to 8 minutes built from three atoms, (10) a FAQ of six questions and answers, (11) an image prompt describing one visual per asset in 25 words or fewer, and (12) a short newsletter teaser pointing at one other asset. Match each format's conventions rather than pasting the same text everywhere. Do not invent statistics, studies, quotes, or numbers. Do not open with filler urgency or a claim that the topic matters now more than ever. If an atom needs a fact I did not supply, write [CONFIRM FACT].
Fill in these details
- Atoms: Your extracted list from stage one, kept as a numbered list you can reference.
- Platforms: Where each asset actually runs. A thread for LinkedIn and a thread for X are not the same object.
- Voice: Your normal writing rules, stated once here rather than repeated per asset.
- Numbers: Lengths you can actually deliver. A twelve-asset plan with no time budget is a plan you abandon.
Why duplication is the default failure
When the same atom feeds four assets, the audience sees the same idea four times in a week and the channel starts to read as one thought repeated. The map above prevents this by assigning one atom per asset, and by requiring you to say what makes each one worth its platform's attention.
The other failure is platform blindness. A caption, a thread and an email are different reading acts. The thread rewards a strong first line and structure. The carousel rewards one idea per slide with almost no text. The email rewards a reason to open it. If the same paragraph appears in all three, at most one of them is doing its job.
Match each asset to its platform
Short video scripts from one atom
A 90 second script needs a single claim and a demonstration or a consequence. Ask for the on-screen element explicitly, because a short video without one is a voiceover with nothing to watch.
Threads and posts
Ask for the first line separately from the body. The first line is read in the feed, and the body is read only after the click or the follow. Optimizing the body first is the usual mistake.
Questions and FAQs
This is the format most often worth doing because it converts objections into content. Take the questions your audience asked in the last month and answer them with one atom each.
Worked example: one interview into twelve assets
A 52 minute interview with a freight company owner about customs delays for small importers becomes twelve assets from eighteen atoms. Four atoms went unused because they needed context the audience would not have.
| Asset | Atom used | What made it platform-specific |
|---|---|---|
| LinkedIn post | The three-document check before shipping | Opens with a concrete failure, no career summary |
| Carousel | The five customs fields that cause stops | One field per slide, three words per slide |
| Thread | The story of the delay that cost a customer a week | Story structure, one beat per post |
| The checklist the guest gave away | The reason to read is a usable checklist | |
| Blog section | How the owner trained staff to talk to angry customers | The only asset that runs to 400 words |
| Quotes | Three lines about being underestimated as a small firm | Each with the sentence before it for context |
| Video, 90 seconds | The 30 second check that catches most stops | Demonstrated on a document, not narrated |
| Video, 6 minutes | Three atoms combined into a how-to | Needs the wider edit, flagged in the plan |
| FAQ | Six questions about small importer duties | Each answer one atom, no invented fees |
| Image prompts | One visual per asset | Named focal point, no AI style mashups |
| Newsletter teaser | Points at the checklist | Sends the reader to the email asset |
| Question set for the next post | Six audience questions | Becomes the next episode's outline |
The four unused atoms are the useful part of that table. Knowing what you could not repurpose tells you which parts of the next episode to expand deliberately.
Check the twelve before you schedule
1. Confirm each asset uses a different atom
Lay the twelve drafts side by side and check for repeated ideas. If two assets could swap atoms without anyone noticing, one of them is not worth publishing.
2. Read the first line of every short asset
In a feed, the first line is the whole pitch. Rewrite it if it starts with context rather than the point.
3. Verify every factual atom
Extraction preserves what was said, including things that were wrong on tape. Check any number, fee, duty rate or timeline against your own records or the source's own materials before publishing.
4. Remove the cross-references that made sense in the episode
An atom taken from the middle of a conversation often carries "as I mentioned earlier". In a standalone post that phrase confuses the reader. Strip it.
5. Schedule by independence
Post assets in an order where each one stands alone, and check that no asset depends on having seen another. If one does, mark it as part two rather than letting it run cold.
Common repurposing prompt mistakes
- Asking for twelve posts in one pass. You get one post twelve times, and you cannot tell which draft failed.
- Ignoring atom count. If the source supports five assets, plan five. Padding shows.
- One text for every platform. Format conventions are the reason a repurposed post outperforms a copy.
- Skipping verification. Extraction preserves errors as faithfully as it preserves facts.
- Publishing the same week. Space the assets, or the repetition is visible to the people you are trying to reach.
Frequently asked questions
How many posts can one video realistically become?
It depends on how many distinct atoms it contains, not on the target number. Let the extraction prompt report what the source supports. A 50 minute conversation with one argument may give you four honest assets.
Should I repurpose with AI or edit by hand?
Use AI for the drafts and volume, and hand-check every claim, first line and platform convention before publishing. Repurposed content is the easiest place for a factual error to reach an audience that trusts you.
Does repurposing get flagged as duplicate content?
Platforms treat repeated material differently and change their rules. Repurposed assets that are genuinely rewritten and platform-native are safer than near-copies of the original text. Check the current guidance for each platform rather than relying on general advice.
Repurpose the last thing you already published
Take your most recent episode or video, run the extraction prompt, and publish whatever number of assets the atom list honestly supports. Schedule the first two this week and keep the rest as a queue rather than a plan. For more content production workflows, browse the content creation prompt collection.