You recorded a 45-minute video last week. It went on YouTube. Then nothing. No blog post, no tweets, no newsletter. Not because you’re lazy — because turning that video into seven other formats manually takes almost as long as recording it did.
That math never works. But there’s an ai content repurposing workflow that changes it entirely. One video in, ten posts out, 90 minutes total. Here’s the system I actually use.
Why Most Creators Skip Repurposing (It’s Not Laziness)
The real bottleneck isn’t motivation. It’s time.
Manual repurposing — watching the video back, taking notes, writing a blog post, trimming quotes for LinkedIn, threading insights for X, adapting the best bits for a newsletter — runs 6-8 hours per piece of content. You already spent 4 hours creating the original. Nobody has another 8.
This is why content repurposing tools exist. But here’s where most creators get stuck again: they download Castmagic, sign up for Descript, open ChatGPT, and now they have three tools and zero workflow. New tools without a system just create a new flavor of overwhelm.
What you actually need isn’t another tool. You need a sequence — four steps in a specific order that turn one transcript into publish-ready content for every platform that matters.
The 4-Step AI Repurposing Workflow
Step 1: Extract the Transcript
Everything starts with a clean transcript. Three options, depending on what you already use:
- Castmagic ($15–49/mo) — best for podcasters. Upload audio or video, get a transcript plus AI-generated show notes automatically. Worth it if you publish weekly.
- Descript (free–$24/mo) — best if you also edit your video. The transcript is a byproduct of the editing workflow you’re already in.
- YouTube auto-captions — free, surprisingly decent. Download the .srt file and run it through Claude or ChatGPT with a cleanup prompt to strip timestamps and fix errors.
Pick one. The goal is a clean text transcript — everything downstream depends on it.
Step 2: Transform with a Master Prompt
This is where the ai content workflow gets interesting. Instead of writing separate prompts for each format, use one master prompt that extracts everything at once.
Feed your transcript to Claude or ChatGPT with a single prompt that returns: five key insights, one blog post angle with a headline, three Twitter/X hooks, two LinkedIn angles, and a newsletter subject line with an opening paragraph. One prompt, one pass. You now have raw material for every format. The advanced system prompts I’ve written about work well as a foundation here — add format-specific output requirements and the model delivers structured results you can actually use.
Step 3: Format for the Platforms That Matter
Most guides tell you to repurpose content automatically across 15 platforms. Don’t. Four formats cover 80% of the value:
- Blog post (600–1200 words) — expands one insight into a searchable, linkable asset
- Twitter/X thread (6–8 tweets) — the punchiest hook, structured for engagement
- LinkedIn post (150–300 words) — professional angle, one actionable takeaway
- Newsletter segment (200–400 words) — personal angle, drives subscribers back to the full piece
Five mediocre posts across twelve platforms lose to four strong posts where your audience actually lives. Run platform-specific follow-up prompts on the master extraction output — not on the raw transcript. The extraction already did the thinking.
Step 4: Distribute
Schedule through Buffer or native platform tools. This step takes 10 minutes and doesn’t need AI. Keep it boring.
The system is Extract → Transform → Format → Distribute. Every step feeds the next. Nothing gets created from scratch after the first transcript.
3 Prompts You Can Copy Right Now
The system is only as good as the prompts driving it. Here are three I use weekly — they work in Claude, ChatGPT, or Gemini.
The Master Extraction Prompt:
“Here’s a transcript from a [length]-minute [video/podcast]. Extract: (1) the 5 strongest insights, (2) one blog post angle with a headline, (3) 3 Twitter/X hooks under 280 characters, (4) 2 LinkedIn post angles, (5) one newsletter subject line + 2-sentence opening. Return as a structured list.”
The Blog Post Prompt:
“Take this insight: [paste from extraction]. Write a 600–800 word blog post with a hook opening, 3 supporting points with specific details, and a CTA. Voice: [conversational/authoritative/casual]. No filler sentences.”
The Thread Prompt:
“Turn this hook into a 6–8 tweet thread: [paste hook]. Tweet 1 is the hook. Tweets 2–6 are evidence or examples. Final tweet is a CTA. Each tweet under 280 characters. No hashtags.”
These look simple. They are. The power isn’t in prompt complexity — it’s in feeding them structured extraction output instead of a raw transcript. That single change makes the output publishable with light editing instead of a full rewrite.
But “light editing” isn’t “no editing.”
The Honest 90-Minute Breakdown
Here’s what 90 minutes actually looks like:
- 30 minutes — upload, transcribe, run the master extraction prompt, scan the output
- 30 minutes — run format-specific prompts and review each piece
- 30 minutes — edit for your voice, fix anything time-sensitive, schedule
That last 30 minutes isn’t optional. Video to blog ai gets you 80–85% of the way there. The remaining 15–20% — your specific voice, current references, technical nuance — still needs you. If you want better raw output from the AI step, solid prompt engineering techniques make a measurable difference in how much editing you’re left with.
But compare that to the alternative. Six to eight hours of manual work compressed into 90 focused minutes. Same output count. Better consistency, because the system runs the same way every time.
That 45-minute video you posted last week? It’s still sitting there, full of content you’ve already created but haven’t used yet. Pick one video from the last 30 days. Run it through Step 1 today. You’ll have ten pieces of content before lunch.