Big Idea

The argument in one line.

A five-step Claude workflow, clip, caption, plan, design, animate, turns a raw podcast into a fully designed, animated vertical reel for about $25 in generation credits and roughly 20 minutes of hands-on work.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

A podcaster or long-form creator who wants to turn episodes into visually rich short-form reels without hand-animating every scene.
A video editor curious about using Claude alongside an AI generation platform (Higgsfield) to produce on-brand motion graphics for social clips.
Someone already comfortable with basic clipping and captioning who wants the next layer: automated scene design and animation planning.

SKIP IF…

You want AI to write or research content rather than edit existing footage.
You don't have an existing long-form source video to clip from.
You're looking for a free workflow. This one runs on paid generation credits plus a paid course for the actual trained skill files.

TL;DR

The full version, fast.

The video walks through a five-phase workflow for turning a long-form podcast into a designed, animated short-form reel using Claude alongside an AI generation platform, a captioning tool, and a transcription service. Claude first reads a timestamped transcript and extracts candidate clips with a custom 'clipper' skill, then trims the chosen clip from the source video. A captioning tool adds subtitles. A second Claude skill plans which moments stay as plain talking-head footage versus becoming a motion-graphic scene, then a 'designer' skill prompts an image model to generate matching vertical designs, and an 'animator' skill prompts a video model to animate them. The creator reports about an hour of mostly unattended AI processing, 20 minutes of hands-on work, and roughly $25 in generation credits per reel.

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 – 00:47

01 · The five-phase promise

Cold open stating the goal: turn a raw podcast into a viral vertical reel using five phases, clip, caption, plan, design, animate, each handled by Claude or a paired tool.

00:49 – 02:53

02 · Phase 1: Clipping from a timestamped transcript

A transcription service generates a timestamped JSON transcript of the source podcast, which Claude reads using a custom 'short-form clipper' skill to extract candidate clips, then uses the same timestamps to trim the chosen clip out of the video.

02:55 – 03:45

03 · Phase 2: Captions

The trimmed clip is captioned in a dedicated captioning tool: a minimal style, light font weight, reduced scale, positioned in the lower third.

03:47 – 05:58

04 · Phase 3: Planning the edit

A second Claude skill, trained on past scripts paired with their exact planned animations, decides scene by scene whether each moment stays as plain A-roll or becomes a generated graphic, explicitly building in pacing 'rest' beats.

06:00 – 06:59

05 · Phase 4a: Generating scene designs

Claude prompts the generation platform's image model to create a still design for each planned graphic scene, using a locked color theme and a design skill trained on an outside mood board.

06:59 – 07:57

06 · Sponsor break: the paid course

Mid-video pitch for the creator's paid editing course, run during the design-generation wait time, covering its modules and a limited-time price.

07:57 – 08:36

07 · Reviewing the generated designs

Walkthrough confirming each generated design matches the plan and holds a consistent palette and style across every scene.

08:38 – 10:43

08 · Phase 5: Animating the designs

A detailed per-second animation prompt is tested on a handful of designs, then fed back into Claude as a reusable animator skill; the resulting animations are reviewed and dragged into the timeline.

10:43 – 11:07

09 · Manual post-production polish

A short manual pass in the editor: an adjustment-layer zoom, sound design in the first five seconds, a subject-tracked crop, and a music bed.

11:07 – 11:36

10 · Cost and time breakdown

The full credit cost and per-phase time spent are tallied: about $25 in generation credits and roughly an hour total, most of it unattended processing.

11:36 – 12:40

11 · The finished reel plays in full

The completed vertical reel, captions, generated scenes, animation, and music, plays start to finish as the payoff.

12:40 – 13:02

12 · Outro and CTA

Closing ask for comments on AI video editing and a nudge toward two more videos.

Atomic Insights

Lines worth screenshotting.

The workflow treats a timestamped transcript, not the video file itself, as the thing Claude reads: without exact time codes in the transcript, Claude can't find or trim clips.
Every custom Claude skill in this pipeline (clipper, planner, designer, animator) was built by feeding Claude the creator's own past editing decisions and having it extract the pattern, not by writing a one-off generic prompt.
The planning step decides scene by scene whether a moment should stay as plain talking-head footage or get replaced with a generated graphic, a pacing decision made explicit rather than left as a default.
Design and animation run as two separate AI passes: an image model generates the still design first, then a video model animates that specific finished design, rather than one model doing both in one shot.
The full reel (8 images, 8 animations) cost just under 500 generation credits, priced at roughly $26 per 500-credit top-up, putting the AI-generation cost near $25 for one finished reel.
Of the roughly hour-long total process, most of the time was unattended AI processing; the creator estimates only 15 to 20 minutes of actual hands-on work.
The design skill was trained on a curated mood board of outside reference images, not on the creator's own past videos, meaning the visual style came from external references rather than the source footage.
A locked two-color theme is specified directly in the generation prompt, which is what keeps AI-generated scenes visually consistent across a single multi-scene reel.
The animation prompt spells out a literal per-second storyboard (second one: object slides in, second two: text pop-fades in word by word, second three: edge elements animate) rather than a vague style description.
Captioning happens before the creative planning pass in this pipeline, so caption placement and style are locked in before the AI ever decides where graphics go on screen.

Takeaway

Five Skills Turn a Podcast Into a Designed Reel

AI EDIT PIPELINE

Turning a raw podcast into an animated vertical reel comes down to five separate AI-driven decisions, one per production stage, rather than a single do-everything prompt.

02Phase 1: Clipping from a timestamped transcript

Claude can't watch raw video for clipping decisions here; it works from a timestamped transcript of the podcast instead, and that transcript's time codes are what let it later trim the actual video file.
A custom clipping skill was built by uploading the creator's own past clip choices and reasoning to Claude, not by hand-writing a generic clipping prompt.
Once a clip is chosen, the same transcript timestamps line up against the source video's timestamps to cut the clip out automatically.

03Phase 2: Captions

Captioning is treated as its own discrete phase between clipping and creative planning, using a dedicated captioning tool rather than doing it inside the video editor.
A caption style gets chosen for legibility on vertical video: minimal styling, a light-weight font, reduced scale, and placement in the lower third.

04Phase 3: Planning the edit

A second custom skill decides, moment by moment, whether a section of the transcript should stay as plain talking-head footage or be replaced by a generated graphic scene.
That planning skill was trained on roughly ten full past scripts paired with the exact animation chosen for every sentence, described as thousands of sentence-to-animation examples.
Sections deliberately kept as plain footage to 'give the viewer some rest' are a pacing choice, not a default; wall-to-wall motion graphics would fatigue a viewer.

05Phase 4a: Generating scene designs

Scene designs are generated with an image model at high resolution, prompted with a locked two-color theme so every scene in the reel matches.
The design skill's training source was a curated mood board of outside reference images, not the creator's own footage, meaning the visual style comes from external references rather than the source video.

07Reviewing the generated designs

Designs get reviewed scene by scene against the plan before animation starts, catching mismatches while it's still a static image and cheap to redo.
A consistent palette and typographic style across every generated design is what makes a multi-scene AI-generated reel read as one coherent piece rather than a grab-bag of styles.

08Phase 5: Animating the designs

Animation is a separate generation pass from design: a video model animates an already-finished still image rather than generating video from scratch.
The animation prompt is a literal second-by-second storyboard, not a vague style request, and that detailed prompt, once proven on a handful of designs, gets fed back into Claude to become a reusable skill.

09Manual post-production polish

Even a fully AI-generated edit still gets a manual post-production pass: a zoom animation, sound design in the first five seconds, a tracked crop, and a music bed.
Final creative judgment, like cutting a slow section or adding sound design, is deliberately reserved for a manual pass after the AI outputs are dropped into the timeline.

10Cost and time breakdown

The full reel's generation cost (8 images, 8 animations) came in just under 500 credits, priced at roughly $26 per 500-credit top-up, putting the AI cost near $25 for one finished reel.
Of roughly an hour of total elapsed time, most was unattended AI processing across the five phases; actual hands-on work was estimated at 15 to 20 minutes.

Glossary

Terms worth knowing.

A-roll: The primary, unedited talking-head or interview footage kept as-is in a scene, as opposed to being replaced by a graphic or motion-design cutaway.
Higgsfield: An AI generation platform used in this workflow for both still-image scene design and animated motion graphics, via separate underlying models for each step.
Seedance 2.0: The video-animation model used inside the generation platform to turn a finished static design into a moving short-form scene.
Claude skill: A reusable, trained instruction set that Claude loads to repeat a specific creative task, built by feeding Claude examples of past human decisions rather than writing a generic prompt from scratch.

Resources

Things they pointed at.

00:49toolElevenLabs (Speech to Text)

02:55toolSubmagic

06:00toolHiggsfield ↗

06:00toolGPT Image 2

08:38toolSeedance 2.0

06:59productUltimate Editors 2.0 / AI Editing Fundamentals Masterclass ↗

Quotables

Lines you could clip.

00:46

“What often looks like discipline today is just the echo of someone's past obsessions.”

a tight, standalone philosophical line with no setup needed, also doubles as the source podcast's own hook→ TikTok hook↗ Tweet quote

12:30

“What once obsessed me has now just simply become me.”

clean thesis line that resolves the personal story arc used as the demo footage→ IG reel cold open↗ Tweet quote

11:23

“20 minutes of real work and $25 for a viral podcast reel like this one.”

concrete, quotable proof-of-cost line that summarizes the whole video's pitch→ newsletter pull-quote↗ 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.

metaphorstory

In this video, I'm going to show you how you can edit viral podcast reels like these using only AI. Even if you've never edited anything with AI before, don't worry. This tutorial is a complete A to Z guide.

I'll show you how to take a raw podcast like this one that was made by Chris Williamson and turn it into this viral Instagram reel using the five -phase podcast reel workflow. Phase one, Claude will watch through our podcast and extract high -quality short -form clips. Phase two, we'll use Submagic to create high -quality captions for our video.

Phase three, Claude will analyze the script. and plan out amazing animations. Phase four, Claude will then prompt Higgs Field to create stunning designs like these for all the clips.

And finally, phase five, Claude will turn all these designs into high quality animations using Higgs Field. At the end of the video, I'll even show you how long this edit took and how much it cost. But first, it all starts with phase one, clipping.

First, it all starts here, with this raw podcast. The first thing we need to do is clip a high -quality short -form clip out of this long -form video. And traditionally, this means watching through the entire video to find the best hook, then placing clips sequentially to formulate a story out of that hook.

But with Clod, we no longer need to do that. Here's how it does the clipping for us. First, Clod needs to be able to watch our podcast, so we feed it a timestamp transcript.

Head over to Eleven Labs, click on Speech to Text, and then Transcribe Files. And since my podcast is on YouTube, I'll just copy the link and paste it. it here.

11 Labs generates a transcript of the full podcast, but this doesn't have the timestamps. So click on this plus sign and generate subtitles for the video. Now you've got timestamps.

Finally, click on export and export as a JSON. A JSON file just shows what is being said in our video and when it is being said. Remember that because those timestamps are really about to matter.

Now I pull up Claude, upload this file into the prompt and tell it to load up a skill called the short form clipper skill and use it to extract five high quality premium short form clips from the script. While it's running, let me take 20 seconds to explain how this skill actually works.

In Ultimate Editors, we have courses on clipping, where I explain how to clip videos like these out of long -form podcasts. So we uploaded them to Claude and trained it on clipping. On top of this, I clipped out multiple clips from a podcast myself and gave Claude a detailed breakdown of why I chose those scripts and why I structured each clip that way.

Claude studied exactly how I clip and turned it all into the short -form clipper skill. And just in case you want these skills for yourself, we're putting all of them inside the AI Editing Fundamentals Masterclass in Ultimate Editors, where you're going to get to learn how to edit viral videos with AI and get access to 12 high quality video editing skills for Claude.

The link is down below. Now that Claude has finished processing, it gives us five high quality clips out of this transcript. I read through all of them and my favorite was clip three.

It uses a relatable topic as a hook with an unexpected reframe. Then it tells a story, jumps in time to an example and wraps up with an outcome. This was perfect.

Now to cut it out, I just uploaded my podcast over to Claude and told it to trim this clip out of the video. It lines up the timestamps on our script with the timestamps of our video. and cuts our clip out of the main podcast.

And now we end up with a high quality short form clip that actually works well, which brings us to the next phase. Next up is phase two, generating the captions. Once you have a clean, high quality clip, it's time to generate the captions.

There are many tools that could generate high quality captions for you. Descript, Submagic, Opus. In this case, I'm going to be using Submagic.

So head over to your browser of choice and load up Submagic. Click on the create project in the top right and then click on upload from computer and upload the clip that we just made. Give it a second to load and then you're going to see this.

Here, click on style and select the style of the subtitles. This is entirely up to you, but I personally always recommend that you go with something clean and minimal. especially for these kinds of edits.

I love the Kelly 3 style, so I'm going to click on it, but I'm also going to make some changes. First, I change the font type here to Poppins and choose the light font weight. Then I scale it down to 30.

And finally, I move it down to the lower thirds of the screen. Now, once your captions look good and you're satisfied, click on export and download the video. Now we've got a high quality clip subtitled in no time, which brings us to phase number three.

The next phase in our workflow is planning out the edit. We need to plan out what needs to be shown on screen in order to make this video go viral. To do this, I'm going to pull up Claude and prompt it this.

I want you to now plan this full edit for optimal virality and viewer retention. To do this, use the podcast clip planner skills that you've been taught and I'll let it run. And of course, while it's running, I'll take the next 20 seconds to explain how this podcast planner skill works.

We have multiple courses in Ultimate Editors that break down how I plan animations for videos. So we fed all of those to Claude and told it to study them. Then I fed it 10 full scripts and the exact animations that we planned for each sentence that's thousands of sentences and thousands of animations planned for them which was more than enough for cloud to study how planning edits work using all this i prompted cloud to extract the content psychology behind how we planned our edits and what we need to show on screen and it created the editor brain skill that we're using here again if you guys want access to all these skills they're all inside the ai editing fundamentals course in ultimate editors the links down below but now that it's done take a look at this for the hook this is what chris says what often looks Looks like discipline today is just the echo of...

someone's past obsessions. Claude planned to keep this as a role to help the viewer connect to the speaker. In the next clip, Chris says this.

I started going to the gym when I was 18 because I was obsessed with gaining muscle. So Claude planned a scene focused on a dumbbell design. The clip after this was this.

Skipped nights out partying to stay in my bedroom and read the misc forums on bodybuilding .com. And Claude planned a form and a TV scene to show what he's talking about. The section after that was the time jump.

Claude planned to leave this as a role to give the viewer some rest. The next section starts with this clip. and Claude planned a fossilized scene for this part.

Then in this clip, Claude planned an orbit scene with all these subjects. After that, Chris says this. What once obsessed me has now just simply become me.

and Claude chose to put a conclusive scene like this. For the section after it, Claude decided to leave it as a roll to give the viewer some rest again. And then Chris comes back with this final statement to which Claude finishes the video with this as the last animation.

As you can see, this edit is planned out like a professional video editor would. And if we follow this plan, you can bet the video is going to turn out so high quality that it keeps our viewer hooked all throughout. And now that we have the raw clip and captions and a plan on what needs to be created, it's time to create it, moving us straight into phase four.

Phase four is the design phase. In this phase, we're going to be turning our plan into high quality designs. To do this, I prompt Claude, create high quality scene designs based on the plan you just created.

Use the podcast designer skill for this and follow a white and orange color theme. Create the designs for every scene in Hicksfield using GPT image 2 at 2K resolution. And then I let her run.

The real secret to getting good designs is the podcast designer skill. What we did here is we went over to Pinterest and gathered hundreds of inspiration of good white vertical designs. All kinds of different styles and colors.

Then I uploaded all... of this to Claude and told it that this is what I want the designs for the short form podcast animations to be like. I told it to study the colors, the design patterns, the psychology of the layout, and everything.

At the end, it created the podcast edit designer skill, a skill capable of generating high quality designs for the vertical style in all kinds of different colors. Look, if you're new to video editing and you're finding it hard to keep up with all this stuff, the new editing styles, the complex programs, and now AI video editing, don't worry, this is completely normal.

This is exactly why we decided to build the Ultimate Editors 2 .0. program. This step -by -step program to teach beginner editors the editing programs, the viral editing styles, AI video editing, and how to make $2 ,000 a month as a video editor.

Inside, you're going to get access to the Premiere Pro and After Effects masterclasses. You'll also get access to the viral editing style breakdowns, where we'll teach you how to create the minimal animation style, the SaaS motion design animations, and even the viral talking head style from scratch. On top of all of this, we just released the AI editing fundamentals course, where we'll teach you how to create animations like Imangaji and Vox from scratch using only AI.

And you'll even get access to the 12 quality video editing skills that we spent two months creating for Claude. After all of this, you'll even get access to the $2 ,000 per month editor blueprint, where I'll show you the secrets to making $2 ,000 per month as a video editor in 2026.

Editors have been killing it with our program. They join as complete beginners, learn to create stunning animations, and they've been making thousands of dollars every month with video editing. If that sounds interesting, click the link down below and join Ultimate Editors.

We're going to be raising the price soon, so make sure to join while it's still $97 a month. Now, after around 10 minutes of processing, Claude had created all the designs.

So let's walk through each one and confirm that it looks good. For the hook, we left it as A -roll, so it's all good. For the first section, it created this dumbbell and text animation, which looks awesome.

For the clip after it, it generated this TV scene. The section after that was left as A -roll, so no need to worry about it. Then for the part after that, Claude created a fossilized dumbbell design, which was super cool.

For the section after that, Claude generated the orbit design with all the subjects. It looks great. For the belief section, Claude created this carved design, which looks awesome.

And finally, Claude chose to end this video with this conclusive scene, which just looks perfect. These designs were awesome.

It all followed the color palette and everything looks premium. Now this brings us to our final phase, phase five, generating the animations. This is where we bring these designs to life by turning them into high quality animations.

This is where I prompted Claude, now using the podcast animation skill, I want you to animate all of these designs into high quality short form minimal animations for my podcast edit. Use the animation logic from the podcast animation skill and generate them using Seedance 2 .0 at 1080p with no music and just sound effects.

And of course, while this is running, I'll finally break down the podcast animation skill that's responsible for generating these stunning animations for our video. Before creating the video, I ran the design skill to make sure everything is working. generated designs, I brought them into Hicks Field in Seedance 2 .0 and prompted it this.

The animation starts with just a white background. In second one, the object slides in in a realistic stone. In second two, the text has a pop -up blur fade -in animation that animates word by word.

In second three, the designs around the edges of the graphics animate in. All animations are quick and smooth and seamless, not rigid and bouncy. I was able to create stunning animations using that prompt.

So then, I did that to another three designs, giving it every detail it needs about the animation. With four high -quality animations created and prompts available, I gave it all to Claude to study and told it to study exactly how I prompt Hicksfield to generate high quality animations like these.

So it studied them and it turned them into the podcast animator skill that understands how to animate high quality designs like these. Now after processing for 15 minutes, Claude was done creating all the animations using our podcast animation logic. So let's take a look through them and make sure the quality is good.

For the first scene, the dumbbell slides in and the text writes out. It all looks perfect and notice how the timing and flow is all on point. For the second scene, the TV slides in with a little bounce and the text text all writes out.

This all looked awesome. The third scene the dumbbell drops into the stone and the text writes out smoothly. This looks perfect.

The fourth scene we see a clean draw out of the lines and a pop -up animation for the icons. For the fifth scene we see a smooth slide in and a write out which just looks clean. For the final scene the stone slides in from the camera and the text writes out as the dumbbell drops which all looks awesome.

It's safe to say all the animations are on brand and complete. So now to finalize everything all we have to do is drag the animations into the edit and finalize it. Let's quickly do that.

First I launch my and then I dragged each animation into exactly where Claude planned for it to be. This was easy and simple.

Finally, I wanted to add a final touch for myself. So I spent five minutes doing some post -production to make this edit pop. First, I added an adjustment layer and created a zoom out animation at the beginning.

Then I added some risers and whooshes to hook the viewer in the first five seconds. And I felt like this first section was a little slow. So I cut this part here and I tracked the subject to create a cool effect.

Finally, I added a good song to the edit and everything came to life. Now, before I show you the final edit, remember at the beginning, I promised to break down exactly how much this cost and how long it took. Here we go.

For the cost, this entire edit took 8 images at 7 credits each and 8 animations at 55 credits each. That's just under 500 credits total. And Higgs Field has a plan where you can top up for 500 credits for around $26.

Which means this entire premium podcast reel cost around $25. As for the time, the clipping took 12 minutes, the captions took 5, and the planning took 8. The design took 10, and the animations took around 16.

And my final touches took around 15 more. This is all just over an hour in total. But most of that was clawed processing in the background while I did other things.

The actual work I put in was around 15, 20 minutes. 20 minutes of real work and $25 for a viral podcast reel like this one. And this, ladies and gentlemen, after the five -phase workflow is the final edit that we end up with.

What often looks like discipline today is just the echo of... someone's past obsessions. I started going to the gym when I was 18 because I was obsessed with gaining muscle and I researched protein shake formulations and dreamt of going to Gold's Gym in LA and skipped nights out partying to stay in my bedroom and read the MISC forums on bodybuilding .com.

The most autistic excuse for not going on a night out ever. Nearly 20 years later, I'm still training. Not because I'm that disciplined or motivated now, but just because an old obsession of mine fossilized into my identity.

This is true. for my meditation habit and my research for podcast guests and the productivity systems that I've got and my desire to build businesses. What once obsessed me has now just simply become me.

This is a quiet reframe that people basically never say out loud. Discipline sometimes isn't the starting point. It's just the residue.

Ladies and gentlemen, thank you guys so much for watching all the way till the end of the video. If you guys enjoyed this, let me know your thoughts on AI video editing in the comments down below. Do you think it's too expensive?

Do you think it was fast? And do you think this is the future of video editing? If you guys want to see more banger viral AI tutorials like this one, comment down below and like this video so you can see more of these.

And in the meantime, feel free to check out one of these two videos over here. I hope you have an amazing day and we'll catch you in the next video.

The Hook

The bait, then the rug-pull.

The creator opens with a promise: an A-to-Z path from a raw long-form podcast to a fully designed, animated Instagram reel, using only AI, no manual after-effects work. What follows is a five-phase pipeline where Claude makes the creative decisions at each stage while a transcription service, a captioning tool, and an AI generation platform do the execution.

CTA Breakdown

How they asked for the click.

VERBAL ASK

06:59product

“click the link down below and join Ultimate Editors... we're going to be raising the price soon, so make sure to join while it's still $97 a month”

Mid-video sponsor break inserted during the design-generation wait time, covering course modules, a $2,000/mo pitch, before/after student testimonials, and price-increase urgency.

MENTIONED ON CAMERA

06:00toolHiggsfield ↗

06:59productUltimate Editors 2.0 / AI Editing Fundamentals Masterclass ↗

FROM THE DESCRIPTION

PRIMARY CTAWhere the creator wants you to go next.