The advice "repurpose your content across platforms" has been around forever and almost nobody actually does it. The reason isn't motivation. It's that repurposing requires having the source material in one place where you can actually read it, plus a fast way to reshape it for each platform's specific rhythm. Most creators have neither. This guide is about putting the source material into one vault and using an agent to do the reshaping. The piece gets published once. The platform-specific versions get drafted from the same source. You stop writing five things from scratch and start writing one thing five ways. ## What repurposing actually requires The operational truth: you can't repurpose what you can't find. You also can't repurpose if reshaping is slow enough that doing it fresh feels easier. Both problems get solved by holding the source material and the platform conventions in one place that the agent can read. The shape that holds: - A **canonical source page** for each substantive piece — a long post, a podcast episode, a video, a presentation. - **Platform conventions** documented as templates the agent reaches for. - **Derivative pages** linked back to the source, one per platform output. - **A reusable pattern** — drop the source in, ask the agent for the platform versions, edit, schedule. When the source and the templates and the platform versions all live in the same vault as plain markdown, the friction drops far enough that repurposing actually happens. The upstream half of this loop — what gets scheduled and when — is in [how to build a content calendar from your notes](/guides/creatives-content/content-calendar-from-notes/). ## The canonical source page Every substantive piece you publish has one canonical source page in the vault. For a long-form post, that's the post itself. For a podcast episode, it's the page that holds the audio, the transcript, and the show notes. For a video, the script or the transcript and the notes. For a talk, the slides as a PDF and the transcript if you have one. PDFs auto-convert to markdown via docstrange when you drop them on a page, so even a slide deck or a guest-author article becomes searchable text the agent can quote from. Audio recordings transcribe with speaker diarization, so an interview shows you who said what — useful when you're pulling a quote and need to know whether it's you or your guest speaking. The canonical source page is the thing all repurposed versions descend from. You shouldn't ever be repurposing from a half-remembered version in your head. You should be repurposing from the page. ## Platform conventions — the agent's reference material Every platform has its own rhythm. A LinkedIn post wants a hook in the first two lines and white space between paragraphs. A Twitter thread wants the first tweet to land and the rest to chain. A short YouTube video wants a script that survives a fast read. An email newsletter wants a personal opener and one clear payoff. A podcast clip wants a hook of context plus the moment plus a tease. Most creators have those conventions in their head and apply them inconsistently. A "Platform conventions" page in the vault makes them explicit. One sub-page per platform. Each page describes what works on that platform — the hook structure, the length range, the tone, the formatting conventions, the call-to-action shape if any. Add notes from your own experience as you publish. When you ask the agent to draft a LinkedIn post from a source page, the agent reads the LinkedIn conventions page first. The output respects your specific approach to LinkedIn, not a generic AI template. As your sense of what works evolves, you update the conventions page. The next draft reflects the new thinking. ## The reshaping pass — one source, five outputs The actual repurposing flow: open the canonical source page, open a "Derivatives" sub-page, ask the agent to draft each platform version. For a long-form post that's about 2,000 words, the request looks like this: "From the draft on the source page, draft a LinkedIn post (using the conventions page), a Twitter thread of 6–8 tweets, a short newsletter blurb of 200 words pointing to the full piece, and three pull quotes for image cards. Reference the conventions pages for each platform." You get all four drafts in one pass. Each one grounded in the source piece and shaped to the platform. You edit. You schedule. The reshaping that would have taken an evening takes thirty minutes. The crucial constraint: every draft links back to the source page. So when you publish the LinkedIn version next week and somebody asks where you got the data point, you can find the source paragraph in three seconds. The trail back never breaks. ## Repurposing podcasts — the highest-value case Podcasts are the most under-repurposed format in content. A 45-minute episode contains material for ten posts and most podcasts publish two clips and a show note. The reason is operational — pulling moments out of audio is slow without good transcripts and slower without a way to query the transcript. With the audio in your vault and a speaker-labeled transcript next to it, the workflow shrinks. Ask the agent: "Read the transcript of this week's episode. Pull the five strongest standalone moments — the kind that work as a clip without context. For each, give me the speaker, the timestamp, and a one-sentence hook for the social caption." You get a structured shortlist. You decide which clips to make. The clip-extraction work that nobody had time for becomes a one-prompt job. Beyond clips: "Pull every claim my guest made that contradicts something a previous guest argued. Reference the previous guest's episode by name." Now you have a thread idea — "Maya Singh and Jordan Bell disagree on remote work, here's where" — that emerges from the corpus rather than from a marketing brief. The full podcaster workflow that feeds this — guest dossiers, prep, post-recording notes — is in [AI notes for podcasters](/guides/creatives-content/ai-notes-podcasters/). For show notes, the same source feeds the public-facing version: "Draft the show notes for this episode. Five-bullet structure. Key timestamps. Two pull quotes." The show notes get published with the episode. The episode page in the vault holds the source for everything. ## Repurposing across long timescales The deeper move is repurposing across long timescales — pulling from your own back catalog months or years after the original was published. Most creators don't do this because finding old material is too much work. With the vault, it's a query. "Find every post or episode I've published in the last two years where I touched on parenting and creative work. Pull the most cited paragraphs from each." You get a shortlist. The synthesis essay you've been wanting to write — the one that connects threads from your past work — can actually get written, because the threads are in front of you. Or for evergreen pieces: "Read my five most-clicked posts from last year. For each, suggest a 2026 update — what's changed, what's new, what would I write differently now." You get an editorial calendar of refresh candidates, grounded in your actual archive. ## Personal voice — staying yours, not theirs The risk with agent-drafted derivatives is they all start to sound the same. Generic. Smoothed over. You publishing posts that don't sound like the posts you'd write. The defense is the source material. The agent is reshaping your own writing — the long post, the transcript of you talking, the email you sent. The voice in the source carries through. If you write the source carefully, the derivatives inherit the voice. If the source is sloppy, the derivatives are sloppy in your name. A second defense is the conventions page. Document not just the platform's mechanics but your own specific way of using the platform — the kind of opener you write, the words you avoid, the pattern of CTAs you use. The agent reads that page and respects it. The output sounds like the LinkedIn posts you'd write, not like generic LinkedIn voice. The same template-and-conventions setup is the spine of [drafting emails, proposals, and newsletters inside your notes app](/guides/creatives-content/draft-emails-proposals-in-notes/). A third defense is editing. The agent draft is a draft. You read it. You change the lines that don't sound like you. You cut the parts that feel hollow. The published version is yours, even though the agent did the structural work. ## The compounding return The first month of running this system feels like extra work. You're documenting conventions, building source pages, getting the workflow into your hands. Month two, the work compounds. You publish a post; the four derivatives draft in twenty minutes. Month six, you publish a post and the agent has months of your past work to pull patterns from — your conventions are sharper, your source material is deeper, your derivative drafts need less editing. The system also surfaces what's been missed. Ask: "Look at every long post I've published in the last year. Which ones never got repurposed into a short-form version on any platform?" You get a list of pieces that did the hard work of writing but never reached the audiences that prefer short-form. Now you know what to pull from when next week's calendar has a gap. ## A vault you stay in The system only pays off if you actually use it. The habit is small: every published piece gets a canonical source page. Every Friday, thirty minutes to review the platform conventions and refresh anything that's drifted. Once a quarter, an hour to mine the back catalog for the connecting essay you've been meaning to write. The agent does the reshaping. You do the source writing and the editing. The middle work — finding the moment, drafting the version, threading the references — gets smaller. The piece gets to more places. Your archive becomes a working asset instead of a folder. Try Docapybara free — [sign up](/accounts/signup/), drop in your most recent long post, and ask the agent for the LinkedIn, Twitter, and newsletter versions.