Big Idea

The argument in one line.

A nine-skill Claude Cowork pipeline can transform one transcript into a full week of content across every platform without toggling between tools, hiring anyone, or writing a single post manually.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

A solo creator or personal brand posting on multiple platforms who is bottlenecked by the time it takes to repurpose content.
Someone already using Claude Cowork who wants to push it beyond single-turn chat into a chained automation system.
A creator comfortable with a one-time setup process in exchange for a repeatable one-click pipeline.
Anyone publishing regularly on X, LinkedIn, Substack, and Instagram who currently uses separate tools for each.

SKIP IF…

You are not willing to pay for both Higgsfield and HeyGen subscriptions -- the visual and video branches require them.
You want a fully no-code setup: this system requires reading prompts, providing brand assets, and configuring avatar IDs.
Your content is primarily visual or short-form video where text repurposing is secondary.

TL;DR

The full version, fast.

Most creators are still making one piece of content at a time, toggling between tools for each platform. Draft Loop is a nine-skill Claude Cowork project that takes a single transcript and runs it through five branches automatically: brand voice foundation, written content (tweets, threads, LinkedIn, Substack), visual content (carousels and AI character storyboards via Higgsfield), video (HeyGen AI avatar reels), and a master orchestrator skill that fires all of them in sequence. Setup is a one-time copy-paste process from a Notion template. After that, the only trigger is typing run the full content pipeline.

Free for members

Chat with this breakdown — free.

Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.

Create a free account →

Chapters

Where the time goes.

00:00 – 01:03

01 · Intro + promise

States the no-gatekeeping premise, lists every output the system produces across all platforms, drops a Notion template QR code, and previews the nine-skill build.

01:03 – 01:38

02 · Transcribing inside Claude Cowork

Shows that Claude Cowork can ingest a YouTube URL and produce a clean markdown transcript saved directly to Google Drive. This is the input layer for the entire Draft Loop system.

01:38 – 02:39

03 · What DraftLoop does

Visual flowchart of the system: transcript in, content edit, then nine skills fire across five branches. Emphasizes that most of the build is copy-paste, not manual work.

02:39 – 04:00

04 · Brand voice skill (Skill 1)

Foundation skill that auto-fires before any written content is generated. Setup prompt reads past Claude conversations to extract tone, phrases, and content pillars. Prevents generic-sounding AI output.

04:00 – 06:00

05 · Written branch: Tweets, Threads, LinkedIn, Substack

Skills 2, 6, 7, 8. Demos the output: 15 tweets, 3 threads, 4 LinkedIn posts, 3 Substack essays from one transcript. Tweet/Thread skill runs two passes. Five Lens Framework adds angle variety.

06:00 – 07:30

06 · Visual branch: Carousels + AI character storyboards

Requires Higgsfield connector. Carousel skill generates slide copy and images in the creator style. Character storyboard produces scene-by-scene scripts for AI video generation.

07:30 – 09:11

07 · Video branch: HeyGen AI avatar setup

Requires HeyGen account. Creator selects indoor and outdoor avatars and grabs their IDs plus a voice ID. Setup prompt locks these into the skill permanently. ElevenLabs is optional alternative.

09:11 – 11:35

08 · Running the full DraftLoop pipeline

Skill 9 (master orchestrator). Create a Cowork project called Draft Loop, paste instructions, upload brand assets, enable memory. Type run the full content pipeline. System audits, approves, then generates everything.

11:35 – 12:23

09 · Content machine in action

Live demo output: carousels generating, LB character storyboard shown, HeyGen talking head video created. Final file tree saved to Google Drive covers all formats.

12:23 – 12:48

10 · Final notes + CTA

HeyGen is optional. Higgsfield subscription required. Notion template in description. Teaser for full marketing team build inside Claude.

Atomic Insights

Lines worth screenshotting.

Claude Cowork can ingest a YouTube URL and produce a clean markdown transcript saved to Google Drive -- no third-party transcription tool needed.
The brand voice skill auto-loads before every piece of written content, pulling tone and phrases from past Claude conversations so output sounds like the creator, not generic AI.
One transcript can produce 15 tweets, 3 threads, 4 LinkedIn posts, 3 Substack essays, 5 carousel decks, a storyboard, and a HeyGen video in a single pipeline run.
The five-lens framework is a separate skill that runs different analytical angles on the same transcript to generate content pillars automatically.
HeyGen requires an indoor and outdoor avatar ID plus a voice ID -- the skill asks for all three during setup and locks them in permanently.
Building skills in the wrong order breaks the master pipeline: brand voice must be Skill 1 because it auto-triggers before every downstream written output.
The master orchestrator first runs a content audit and asks for approval before writing anything, letting you catch misaligned angles before generating 15 pieces.
Claude Cowork memory being enabled on the Draft Loop project is what lets the system remember your brand assets, avatar IDs, and carousel style preferences across sessions.
ElevenLabs voice cloning is a drop-in alternative to HeyGen native voices -- the skill is pre-configured to accept either voice ID format.
The system treats your transcript as raw material, not a final product: the content audit step identifies tweet angles and LinkedIn hooks before writing a single word.

Takeaway

How one input becomes ten content formats.

WHAT TO LEARN

The bottleneck in most content workflows is not ideas or skill -- it is the manual labor of reformatting the same idea for every platform, every week.

A transcript is not just a record of what was said -- it is raw material that contains enough signal to generate tweets, LinkedIn posts, essays, carousels, and video scripts simultaneously.
Brand voice is infrastructure, not decoration: without a skill that auto-applies your tone before every output, AI-generated content defaults to generic patterns that sound like everyone else.
The content audit step before bulk generation is what separates useful automation from noise -- approving angles before writing prevents a pipeline from producing 15 pieces that all miss the point.
Chaining specialized skills in sequence is more reliable than trying to do everything in one prompt: brand voice fires first, formats fire second, master orchestrator fires last.
A one-time setup cost -- copying prompts, pasting avatar IDs, uploading brand assets -- pays back in perpetuity because the pipeline runs identically every time with no additional configuration.
The visual and video branches are optional modules: the written branch alone produces enough content for most creators without Higgsfield or HeyGen subscriptions.
Memory being enabled on the project is what makes the automation feel seamless -- the system already knows your brand, avatars, and style without being told again each session.

Glossary

Terms worth knowing.

Claude Cowork: Anthropic desktop application for Claude that supports projects, skills (saved system prompts), memory, and third-party tool connectors -- distinct from the Claude.ai web interface.
Draft Loop: The nine-skill Claude Cowork pipeline that transforms one transcript into multi-platform content automatically.
Skills (Claude Cowork): Saved system prompt configurations in Claude Cowork that auto-activate based on triggers, allowing specialized behaviors to fire without re-prompting each session.
Higgsfield: An AI video generation platform connectable to Claude Cowork via a custom connector, used to generate carousel images and AI character storyboard videos.
HeyGen: An AI avatar video platform that generates talking-head videos using a cloned voice and selected avatar. Requires a subscription and avatar/voice IDs for integration.
Five Lens Framework: A content repurposing skill that analyzes a transcript through five different angles to generate varied content hooks and pillars.
Content audit: The first step the master pipeline runs after receiving a transcript: identifies tweet angles, thread ideas, and platform-specific hooks, then asks for approval before generating anything.

Resources

Things they pointed at.

00:39productNotion template (DraftLoop) ↗

00:58toolClaude Cowork ↗

06:25toolHiggsfield ↗

07:29toolHeyGen ↗

08:16toolElevenLabs ↗

06:32linkHiggsfield + Claude setup video ↗

Quotables

Lines you could clip.

00:00

“There's gonna be no gatekeeping on this video.”

Instant trust-building opener that hooks the AI-content audience tired of vague tutorials→ TikTok hook↗ Tweet quote

02:51

“So it doesn't sound like generic slop.”

Punchy, relatable, captures the core fear of AI-generated content→ IG reel cold open↗ Tweet quote

09:16

“All you have to do is type in one thing and it creates everything.”

The core value prop in one sentence, no setup needed→ newsletter pull-quote↗ Tweet quote

21:25

“Ideas come from God. If you don't execute, it goes to somebody else.”

Standalone motivational line from the AI-generated character storyboard demo→ IG reel cold open↗ Tweet quote

The Script

Word for word.

Read-along

Don't just watch it. Burn it in.

See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.

00:00There's gonna be no gatekeeping on this video. Everyone talks about creating content with AI, but no one's showing you the complete system. I'm talking about a system where you could put your transcript, your meetings, your brainstorming sessions, and it will create tweets and threads, carousels, LinkedIn posts, Substack articles, a whole talking head video with your AI avatar, and a story with your AI character.

00:23All in one tool, one location. You don't have to toggle through all these AI tools. And in this video, we're gonna break it down section by section.

00:31I promise you it's going to be extremely simple, literally a copy and paste. And I'm a walk you through each step from step one of creating the transcript all the way to the complete system where all you have to do is type in one thing and it'll make everything.

00:44And to make it easier and for you to go along with it, I created a notion template that's in the description and the QR code is about to pop up so you could follow along with all the skills and the prompts that we're going to need. This is going to be a Claude Cowork masterclass. So get your Claude open, get your pen and your pad, and let's start with step one.

01:03So step one, we're gonna open up Claude Cowork. Now what most don't know is that you can transcribe a video and audio right inside of Cowork. If you don't believe me, you could put in the prompt of if I gave you a YouTube URL or gave you a video, could you transcribe it?

01:16And it will say yes. And so what we did was gave it a YouTube URL for this example, but you can upload your own video and it made the transcript right here into a markdown file that automatically saves into your Google Drive.

01:33Now that we have the input to run this, now let's build out the system. Now what we are about to create is what I call draft loop, where you're taking your videos, your journal entries, your meetings, and you're turning it into a draft machine where the transcript is now turned into written visual and video content automatically inside a co work.

01:55Now to visually see what we're about to build with Draftloop is that we already have our transcript. Once we have that and we enter in co work, it's gonna go through a content edit. Then we're going to build out everything from the threads to the LinkedIn polls to the carousels and the Heijin AI avatar.

02:14Now overall, we're gonna build nine skills, but I don't want you to be intimidated because majority of this is going to be copy and paste. So I don't want you to feel like manually you're gonna have to do anything. To break it down even simpler, it's only five sections.

02:26We have our foundation, we have our written side, the visual, the video, and then at the end of this video, we're gonna show you how it is simple to just type in one thing and it creates everything automatically.

02:39So the first thing we have to do to set up the system is the foundation which is our brand voice. This is a very important skill to set up first and foremost because this determines our tone, our patterns, our phrases, everything for it to sound exactly like us when we're making our tweets and LinkedIn and Substack posts.

02:56So it doesn't sound like generic slop. Now I'm a set you up for success because the prompts that I'm about to give you is going to read your past conversations and see what your brand voice and the rest of the skills are going to be.

03:09And if you don't have any information in it or if it doesn't have everything is going to ask you questions. So in your Notion page, you're gonna see this setup prompt. And so I want you to copy it, go into Claude Cowork, hit customize, go into new skills, and hit the plus sign, create skills, and create with Claude.

03:29What we're going to do is get rid of this, hit that setup skill right there. So it's going to set it up to ask you questions and go through your past Claude.

03:40But then we're going to get the full brand voice skill. So this is what the full brand voice skill and it's going to get everything that it needs from you as far as your name, your brand, your audience, all of that is going to be in there.

03:53Copy it, you paste it, and now you're gonna hit enter. And literally, we're going to be doing this with all the skills.

04:00So now we're getting into the actual content. So we're gonna build out the threads, the LinkedIn, the Substack, and the five lens framework.

04:08So getting into the written side like the tweets and the threads and LinkedIn posts, let me show you what it can actually create before we set up these skills. So I had put a transcript in and it created these tweets. So it gave me 15 tweets from the transcript.

04:23Right? Then it gave me three threads that I can use. Right?

04:29And this is good for x or threads. It also has my LinkedIn posts all in this document. So it has three different LinkedIn posts as you can see.

04:41I lied, actually four. And last part about the written one is the Substack. It actually made three different essays that I can use and put onto my Substack.

04:50And it saved everything into my Google Drive. Now for the Tweets and Threads, we don't have to do that setup prompt that I showed you earlier. It's very simple.

04:59It's gonna go through a Tweet pass and a Threads pass. So you can have your options of how you wanna go either short or whether you have more of a dialogue into your written content.

05:11So for each of these, we don't need to do the setup prompt. All we have to do is get the actual skill and put it into quad.

05:18So for example, for the tweets and threads, we're gonna go down to where it says the full skill instructions. We're gonna copy that. We're gonna go back into skills and all we're going to do is paste it and press go.

05:31And we're doing that with each of these. With the LinkedIn one, we're just gonna go and copy the instructions.

05:37We're gonna do that for a substack, do the same thing, as well as the five lens framework.

05:43So when it's all said and done with the written section, you'll be able to generate your tweets and your threads and everything that we just did from the written section. But now how do we do the visual?

05:53So remember at the end, we're going to set it up to hit a button, but we can't miss any of these steps. So let's get into the next one. With the visual side, we're gonna make carousels and a storyboard for our AI character.

06:05And so what it's going to do is look something like this. So this is kind of style that I like where it looks like it's being written on paper, has a pen. Also, I had created a storyboard with my LB character so I can set it up that my video can just do all of these different scenes.

06:21Now to make this work, we don't only just need the skill, we need the connectors. And for this to work, we need the Higgs field connector. I did do a video about how to set up Higgs field and Claude, and I'll drop that up top and also in the description.

06:35But to make sure that you have this connection, you're going to go into connectors and it should say something right here with Higgs field.

06:44If not, you're going to have to do add custom connector and I walk you through that all in the video. Once that connection is set up, now we gotta make the carousel and the storyboard skill so that it can automatically do it.

06:57Now for the carousel and the story, we do have a setup prompt. So if I go into the carousel, you'll see step one is going to say use this prompt to set it up.

07:07So we'll just do copy and kind of the same thing go into skills and put this there as well as the actual skill underneath it. And it's gonna be the same thing with the character storyboard.

07:18You have your setup prompt and then you have the actual skill. I got two more steps. Remember, we can't skip anything and this one's going to allow us to scale ourselves and not having to record videos all the time.

07:30So now we're doing the video side and setting up Heijin and Claude to talk to each other. Now first things first to automate this, we gotta go inside of Heijin and pick out the avatar we always wanted to use and the voice we wanted to use. Now the way that I have the skill set up, it's gonna always ask me if I want an indoor avatar or an outdoor avatar.

07:49Why? Because what I wanna do is have a variety of avatars.

07:55So I will pick this one as well as this one right here. And all I would need is the actual copy the avatar ID for both of them.

08:07And we're actually going to do the same thing with voices because inside of HeyGen, it does clone your voice, but I also have it hooked up to 11. So either way you can use the hey Jen voice or you can set up a third party with eleven Labs to clone your voice. And so the same thing you're gonna want to do is you're going to go into the voice that you want and hit get voice ID.

08:31But the important part is to have those IDs because when you have the setup prompt for your skill is going to ask for that. So just replace the voice ID right here with your actual one, the indoor avatar and outdoor avatar.

08:47Now, if you don't have an indoor outdoor, that's fine. This is where you'll just tell it I have one avatar and just put it there and it will adjust.

08:57We're gonna do the same process that we've been, take the setup prompt, then as well as take the full skill and put it inside of Claude. And now, let's put it all together so we never have to worry about doing this over and over again.

09:11Now the last thing that we're going to do is set up this last skill that brings it all together, which is the whole draft loop. Now once this is set up to trigger all that we did, all we have to type in is run the full pipeline. So this is the full skill instruction and we're gonna copy this and create this skill.

09:27Now my suggestion is creating a co work project called draft loop to have this run so it knows the instructions that you want and it knows exactly what you want to do. So as you can see, I already have mine set up, but if we was to make it from scratch, we'll hit new project. We'll go start from scratch, draft loop, have that up top, and then you're gonna paste these instructions.

09:50Now, I'm gonna give you the instructions inside of Notion, but what I want you to do is reverse engineer it so it can customize it to your brand and your content and all the skills that you just created. And what you're going to put into these files, let's say you have a logo, let's say you have your AI character, let's say you have the certain type of carousels that you want it to look like, you're gonna put all of those files right in here And then memory is on and hit create.

10:18So anytime you have a transcript, you're gonna put it right here. And then you say, can you run the full content pipeline?

10:25Hit enter and it's gonna start. So once I add my prompt, it's going to ask me my call to action. It's gonna go through my content audit.

10:34So it figured out the tweets from it. It figured out the threads from it.

10:40The LinkedIn post just the topic wise because this is just the audit. It hasn't written anything out just yet. But it you can see that it's creating everything.

10:49Right? And for the sake of this video, I'm just gonna say yes. But normally I would comb through it and kind of make tweaks for however I want it to be.

10:57So now that I say yes, it's gonna start working. And with co work, it shows you the whole progress. So it already did the content audit, and now it's writing all the written content.

11:07It's gonna do the visuals. It's gonna do the video, and then do the carousel.

11:13And now you can see it's starting to make the carousels, all the images with the text that we already approved. Here is the character

11:22story that we made. I had an idea. I sat on it.

11:26I thought I had more time. Ideas come from God. If you don't execute, it goes to somebody else.

11:31They may not do it as well as you can. Act on what was given to you. And as you can see, it created my talking head video right here.

11:39And hey, Jen, if there's a part you're not sure is gonna work, that's probably gonna work. Hands down.

11:46And the reason is simple. We overthink. Your first instinct was fine.

11:52If you thought, I think this could work, but I'm not sure, post it. It'll probably work.

11:58And it saved all the written content, the images and the videos inside of my Google Drive. Now, this is where AI has truly become a content machine by you just giving it one thing and it creates everything else. Now, Heigen isn't completely required to run draft loop.

12:15Some people don't feel comfortable cloning themselves, but I wanted to show you the process. If you don't want to add that, you can remove it in your instructions. And that is my complete content system called Draft Loop.

12:25I'm gonna give you the Notion page that's gonna be in the description. You do need a Higgs field subscription to run it and a HeyGen subscription to run it, but it's all inside of Claude once you have it all connected. And if you want to learn more about Claude Cowork and building out your whole marketing team, I got that right here.

The Hook

The bait, then the rug-pull.

No gatekeeping. That is the opening promise, and the creator means it: within twenty seconds, she has listed every output the system produces, named the tool, and told you it is one location. Most AI content tutorials show you one trick. This one shows you the whole machine.

Frameworks

Named ideas worth stealing.

01:38model

Draft Loop

Foundation (Brand Voice)
Written Branch
Visual Branch
Video Branch
Master Pipeline Orchestrator

Nine-skill Claude Cowork pipeline that converts one transcript into multi-platform content across all five content types automatically.

Steal forAny creator who repurposes content across platforms and wants to collapse that into a single trigger.

04:00concept

Five Lens Framework

A skill that analyzes a transcript through five different angles to generate content hooks and pillars, ensuring variety rather than repetition across platforms.

Steal forBreaking out of the single-angle trap when repurposing a long video into short-form content.

10:30concept

Content Audit Step

Before generating any content, the master pipeline runs an audit that identifies tweet angles and LinkedIn hooks, then asks for approval. Prevents misaligned pieces from being written.

Steal forAny AI content workflow where you want a human checkpoint before bulk generation.

CTA Breakdown