ع
Start Topics Teams Reference What's new Saved
playbook

The content repurposing engine

Turn one strong asset — a post, a webinar, a launch — into a week of on-brand content across every channel, without inventing new claims.

easy ~45 min
when to reach for this

You make one good thing — a blog post, a webinar, a customer interview — and then it dies after one share. Meanwhile the team needs LinkedIn, X, email, and a newsletter blurb every week. The repurposing engine is a repeatable system: feed Claude one source plus your brand voice, and get a full week of channel-native content that all traces back to claims you already stand behind.

gather this first
  • The source asset as a file — launch-post.md, webinar-transcript.txt, or a pasted-in interview.
  • A brand-voice.md doc: 3–5 voice rules, a do/don't list, and 2 posts you were proud of. This is what makes "our voice" mean something.
  • Your channel list and any hard rules per channel (X ≤ 280 chars, no emoji on LinkedIn, etc.).
the workflow

  1. Prime Claude with the source and the voice

    Open the folder with your source asset and brand-voice doc in Claude Desktop and ask in the chat — no terminal needed. Don't start by asking for posts. Start by making sure Claude has read the raw material and can describe your voice back to you — that catches a bad voice doc before it taints 20 drafts.

    you ask
    Read launch-post.md and brand-voice.md. In 3 bullets, tell me: the single biggest idea in the post, the 3 most quotable lines, and how you'd describe our voice in one sentence. Don't write any posts yet.

    what you get back A short read-back: the core idea, three pull-quotes, and a one-line voice summary. If the voice summary is wrong, fix the doc now — not after it's written a week of off-brand copy.

    This "read it back to me first" move is the highest-leverage habit in the whole system.

  2. Batch the channel-native drafts

    Now ask for volume, but be specific about count, format, and the angle each one takes — vague asks get vague, samey drafts.

    you ask
    Now turn the post into: 5 LinkedIn posts and 8 X posts. Each should pull a DIFFERENT angle from the article — no two saying the same thing. Match brand-voice.md exactly: confident but plain, no exclamation points, no emoji. X posts under 280 characters.

    what you get back 13 ready-to-schedule drafts, each mining a distinct angle (the problem, the proof, the customer quote, the contrarian take, the how-to). All in your tone.

  3. Add the email and newsletter cut

    Same source, different format. Keeping it in one conversation means the email stays consistent with the social copy instead of drifting.

    you ask
    From the same post, write a 150-word announcement email — give me 3 subject-line options, preview text under 90 characters, and one clear CTA. Then a 60-word blurb for our newsletter that links to the post.

    what you get back A tight email with subject-line options and preview text, plus a short newsletter blurb — all consistent with the social batch.

  4. Lay it out as a shippable calendar

    Turn the pile of drafts into a plan you can hand to a scheduler. Ask for one table so it's paste-ready.

    you ask
    Lay all of this out as a 5-day posting calendar starting Monday: what goes out, on which channel, at what time, in one table. Front-load the strongest two posts.

    what you get back A day-by-day table mapping each draft to a channel and slot — your week, planned, in one view.

make it your own
  • Webinar instead of a post: swap step 1 to Read webinar-transcript.txt and pull the 5 best standalone ideas — the rest of the engine is identical.
  • Keep a swipe file: maintain swipe-file.md of lines that landed and paste it in alongside the voice doc, so each batch sounds more like you over time, not less.
  • Save the recipe (Power Track): once the prompts work, store them as a /repurpose custom command (see the Playbook's Capabilities tab) so next week is one command, not four prompts.
watch out for
  • Repurpose one strong piece into many formats — don't let Claude invent new claims or stats you'd then have to fact-check. The whole point is reusing things you already stand behind.
  • Read every draft for claims and numbers before it ships. Claude can state a statistic with total confidence that is simply wrong. You are the fact-checker.
  • Keep one approval gate between the batch and the scheduler. At volume it's tempting to publish straight from the queue. Don't — route the week through whoever owns brand, and send anything with a price, a claim, or a named customer to legal sign-off first. The engine's job is to fill the queue, not to skip the reviewer.
  • If the drafts feel generic, your brand-voice.md is too thin — add real example posts, not adjectives. "Witty" tells Claude nothing; a post that was witty tells it everything.

you'll end up with A repeatable system that turns one asset into ~15 on-brand pieces and a ship-ready calendar in under an hour — every piece traceable to something you already published.

Questions people ask

Will Claude invent new claims or statistics when repurposing?
It can, which is why the rule is to repurpose one strong piece into many formats — not generate new facts. Read every draft for claims and numbers before it ships; Claude can state a wrong statistic with total confidence, so you are the fact-checker.
What do I need before I start?
One source asset as a file (a post, webinar transcript, or interview), a `brand-voice.md` with a few voice rules and example posts, and your channel list with any hard limits like "X under 280 characters" or "no emoji on LinkedIn."
Why does the first step tell Claude not to write any posts yet?
Because having Claude read back the core idea and your voice first catches a bad voice doc before it taints twenty drafts. It's the highest-leverage habit in the system: fix the summary now, not after a week of off-brand copy is written.
My drafts come out generic — what's wrong?
Almost always a thin `brand-voice.md`. Add real example posts, not adjectives: "witty" tells Claude nothing, but a post that was witty tells it everything to pattern-match on.