One Video Becomes Thirty Pieces
I see patients most of the week. I film once. The skill that turns one recording into a week of clips, a carousel, a blog, an email, and a thread, all in your voice.
I see patients most of the week. I don't have time to film every day, so I don't.
I film once, long, and the octopus does the rest. Clips with hooks. A carousel. A blog post. An email. A thread. Captions. One recording, a full week of content.
The trick isn't chopping a video into pieces. It's finding the moments that actually stand alone on their own, and rewriting each one in my voice without inventing a single thing I didn't say. This is the exact skill my content agent runs, and it even flags the moments that were great ideas with a weak delivery, so I know what to reshoot.
The three levels
Film seven separate pieces of content because you think each platform needs something original. Burn out by week three.
Feed one transcript to the skill, get the whole week's pieces mapped to platforms, load them into your scheduler as drafts, approve in ten minutes.
Every long recording gets processed the moment it lands, drafts appear in the scheduler, and next week's performance tells the agent which arms to grow. Publishing runs itself.
The free skill
It maps the 8–15 standalone moments in a transcript, produces five clips with hook titles, a carousel, a blog post, an email, a thread, and captions with keywords, packages them into a weekly schedule, and loads them as drafts for approval.
--- name: content-octopus description: Turns one long-form video into a full week of content — clips with hooks, carousels, a blog post, an email, threads, and captions — from the transcript, in your voice, with nothing invented. Trigger on "repurpose this video", "turn this into a week of content", "content octopus", "make posts from this recording". --- # The Content Octopus One long video is a week of content that most people throw away. This skill takes the transcript of one recording and produces every piece the week needs — each one a real standalone, none of them a lazy chop. ## Before you start (connector-first) Get the transcript from a connected source: the platform's transcript connector, a research API via a connector hub, or a transcription service on the raw file. Manual fallback: paste it. Read the whole transcript first. Then read it again for the moments — the story, the number, the opinion, the mistake. Those moments are the octopus's arms. Ask for two things: where each piece will be published (which platforms), and one example of the creator's own writing so every piece sounds like them. ## Step 1: Map the moments List 8–15 moments from the transcript, each with a timestamp and a one-line label: what makes it stand alone (a claim, a story, a number, a contrarian take). Rank them by how well they'd survive with no context. The top 5 become clips; the rest become written pieces. ## Step 2: The arms (produce all of these) **Clips (5).** For each: start/end timestamps, an on-screen hook title (2 lines, 2–4 words each, withholds the payoff — a title that labels the topic is dead), and a caption with a comment keyword. **Carousel (1).** 7–10 slides from the single strongest framework in the video. Slide 1 is the hook, the last slide is the ask. One idea per slide. **Blog post (1).** 800–1,200 words. The creator's argument, restructured for reading: a real headline, a claim up top, the proof, the how. No "in this video I…" — it's a standalone article. **Email (1).** One story from the transcript → one lesson → one link. Under 250 words. **Thread (1).** 6–9 posts. The first post is the hook (see hooks-that-stop), each post a single beat, the last post the ask. **Short captions (3).** Standalone one-liners from the strongest lines, each with a keyword. **Re-film prompt (1–2).** Moments that were great ideas but weak deliveries: a 30-second script with a new hook so the creator can re-shoot it properly. ## Step 3: The integrity rules - Every fact, number, and claim comes from the transcript. Nothing invented, nothing "improved." Where the creator was vague, keep it vague or flag `[CHECK: …]`. - The voice is theirs. Match sentence length, register, and vocabulary from the sample. - No AI tells: no "delve", no "in today's fast-paced world", no rhetorical "But here's the thing" openers, no bullet soup where a sentence would do. ## Step 4: Package for the week Deliver as a schedule: which piece, which platform, which day, with the keyword for each gated piece. If a scheduling tool is connected (Blotato, Buffer, or similar through a connector), load the drafts in a paused state and hand over the calendar for approval. Never publish. ## Step 5: Close the loop A week later, pull performance on each piece and note which arm worked. Over a month the pattern shows: which moments travel, which formats the audience wants, and what to film next. The octopus learns.
Want the octopus running on every recording?
This is one of the highest-leverage systems we run. Inside the AI CEO Lab I teach how to wire it to your scheduler and let agents run it. Or we build it for your brand.