Full Transcript

·YouTLDR

Claude Code (Free Plan) + YouTube = $77,000/Month

12:28694 summary words · ~3 min readEnglishBy zapiwala aiTranscribed Aug 4, 2026
Analyze another video with Pro30-day money-back guarantee
Summary

You can produce high-performing faceless stick-figure YouTube videos entirely for free by prioritizing voiceover rhythm over visuals and automating bulk image generation using Claude, ElevenLabs, FaziScribe, and Google Flow.

This pipeline eliminates the financial and technical entry barriers for faceless YouTube automation, enabling solo creators to produce 100+ scene videos without paid software or manual design skills.

Section summaries

0:00-2:00

Channel Case Study & Claude Setup

optional

The presenter highlights a viral stick-man YouTube channel that achieved 7.5 million views and 137,000 subscribers in two months using only static images. After addressing the hurdle of expensive subscription tools, the presenter introduces a 100% free workflow using Claude AI's free plan. By entering a single master prompt, Claude generates five viral topic ideas and drafts a full 5-to-15-minute long-form script exported as a text file.

  • Faceless stick-figure content can generate millions of views without complex animations.
  • Claude's free tier can output long-form script files directly using targeted prompts.

Establishes motivation and initial scripting, but the technical workflow begins in the next section.

2:00-4:00

The Voiceover-First Rule

watch

This section explains why traditional AI workflows—generating images first and voiceover second—fail due to visual audio collisions. The presenter establishes that voiceovers must be generated first so natural speech pauses dictate visual scene lengths. Using ElevenLabs' free plan (10,000 monthly credits), the creator generates the full narration track using the realistic 'Raunak' voice.

  • Generating voiceovers first aligns editing cuts with natural human speech rhythm.
  • ElevenLabs free tier provides sufficient monthly credits for long-form channel production.

Teaches the critical structural editing principle that impacts video retention.

4:00-6:00

Audio Pause Extraction via FaziScribe

watch

To find exact cut points, the speaker uploads the audio to FaziScribe.ai using its accuracy transcription mode. FaziScribe identifies every speech pause with precise millisecond timestamps across languages. Pasting these timestamps back into Claude prompts the AI to generate distinct text-to-image prompts tailored for every specific line and timestamp block.

  • FaziScribe auto-detects speech pauses to create precise visual markers.
  • Claude can map visual prompts directly to timestamped transcript blocks.

Crucial step for turning audio timing into a synchronized visual prompt manifest.

6:00-9:00

Bulk Image Generation with Zappy Flow

watch

The video transitions to Google Flow using the Nano Banana 2 image model in 16:9 aspect ratio. To bypass manual generation for 100+ scenes, the creator demonstrates the Zappy Flow Chrome extension. The extension accepts all text prompts at once, executing image generation and auto-downloading files sequentially in the background.

  • Google Flow's Nano Banana 2 model generates high-volume assets without tight limits.
  • Zappy Flow extension automates multi-prompt image batching and downloading in the background.

Shows the exact method for automating the most time-consuming step in AI video creation.

9:00-11:00

Timeline Syncing & Metadata Generation

watch

The presenter imports the voice track and downloaded visuals into a video editor, aligning image duration using the timestamp names embedded in each file. After timeline trimming, the creator returns to Claude to generate viral titles, descriptions, tags, and five thumbnail prompts, which are batch-processed via Zappy Flow.

  • Timestamped image filenames simplify timeline trimming in video editing software.
  • Claude handles post-production metadata and thumbnail prompt generation in the same workflow.

Covers practical assembly tricks and post-production optimization steps.

11:00-12:00

Publishing & Summary

skip

The presenter summarizes the entire zero-cost publishing stack from blank page to published YouTube video. Viewers are directed to master prompts in the video comments and encouraged to view follow-up monetization strategies.

  • Complete YouTube automation pipelines can be operated entirely on free AI tiers.
  • Master prompts are provided in video description and pinned comments.

Standard concluding remarks and call-to-action details.

Key points

  • Voiceover-First Rhythm Principle — Generating visual scenes before narration leads to jarring visual collisions; finalizing audio first allows natural speech pauses to dictate exact scene cut points.
  • Pause-Based Timestamp Extraction — Using automated audio transcription tools like FaziScribe reveals every natural pause down to the exact millisecond across multiple languages.
  • Automated Bulk Prompt Execution — Pairing Google Flow's Nano Banana 2 model with custom browser extensions allows automated background rendering and downloading for 100+ scene images.
  • Timestamped File Timeline Alignment — Downloading generated images with embedded timestamp filenames allows editors to drop assets onto video timelines and align trim points in minutes.
Static images, stick figures with 7.5 million views. Narrator
Voice-over first, scene second, always. Narrator

AI-generated from the transcript. May contain errors.

0:00

It's 2:00 a.m. at night. You can't

0:02

sleep. You're scrolling YouTube, thumb

0:04

moving automatically, not really looking

0:06

for anything. Um and then one thumbnail

0:09

stops you cold. A stick man, a simple

0:12

hand-drawn stick man, 7.5 million views.

0:16

Um you tap it. No fancy animation, no 3D

0:20

characters, no cinematic visuals,

0:24

not even video clips, just still images

0:26

of stick figures

0:27

one after another uh stitched together

0:30

into a full video. Um that's it. Static

0:33

images, stick figures with 7.5 million

0:37

views.

0:38

You go to the channel, 14 videos,

0:40

137,000

0:42

subscribers,

0:43

14.5 million total views,

0:46

uh with first video published just 2

0:48

months ago.

0:50

You sit up. One thought hits you.

0:52

If a stick man channel can do this in 2

0:55

months, I why can't I?

0:57

You open your laptop, you try to build

0:59

the workflow, but you hit a wall at the

1:02

very first step. You don't know how to

1:04

write a script that people actually want

1:06

to watch. And every AI tool that can

1:08

help requires a paid subscription.

1:11

You close the laptop. That stick man

1:13

channel just crossed another million

1:15

views while you're still at zero.

1:17

I know that feeling exactly cuz I was

1:19

sitting in the same place uh until I

1:21

built a complete free workflow that

1:23

creates these videos from script to

1:25

final upload. No paid tools, no design

1:28

skills, just a system that works. But

1:32

before we dive in, there's one thing you

1:34

need. A single prompt, not five tools,

1:36

[music] not a paid subscription, not a

1:39

complicated setup, just one prompt.

1:42

I spent weeks engineering and optimizing

1:44

this

1:46

specifically for this workflow. Uh it's

1:48

in the pinned comment below. Copy it and

1:51

now open Claude AI free plan. That's all

1:54

you need.

1:55

You can see I'm on the free plan right

1:56

now. Paste the prompt and and watch what

1:59

happens next.

2:01

The prompt loads.

2:03

And Claude doesn't just give you one

2:04

idea, it gives you five. Five viral

2:07

topic ideas.

2:09

Each one engineered to explode if you

2:11

follow this workflow completely. Pick

2:14

your number, hit send.

2:16

What comes back will surprise you. Not

2:18

just a script, a complete long-form

2:20

video script like anywhere from 5 to 15

2:23

minutes ready to

2:25

ready to use.

2:26

And it doesn't stop there. Claude

2:28

automatically creates a text file you

2:30

can download instantly. Here's the

2:32

download file. Now, before we move to

2:35

the next step,

2:37

I need to stop you.

2:38

Because what I'm about to tell you next

2:40

is the single most important thing in

2:41

this entire video.

2:43

Most creators follow this workflow,

2:46

write script, create text to image

2:48

prompt for each scenes,

2:50

generate scenes,

2:52

create voice-over, then try to sync

2:54

everything together.

2:56

Sounds logical, right? It's completely

2:58

wrong.

2:59

And if you've been following the same

3:01

workflow,

3:03

this is exactly why your videos aren't

3:05

exploding. Here's the problem.

3:07

When you generate scenes first and

3:09

create the voice-over second, you're

3:11

forcing two things together that were

3:13

never built for each other.

3:15

You're not creating rhythm, you're

3:17

creating a collision,

3:18

and viewers feel that collision even if

3:21

they can't explain why they clicked

3:22

away.

3:24

The correct workflow is the opposite.

3:26

Voice-over first, scene second, always.

3:29

Here's why.

3:30

Look at this voice-over. You can hear

3:32

the natural pauses between sentences.

3:35

That pause at 1 second 12 frames.

3:38

That's where scene one ends.

3:40

The next pause at 3 seconds 18 frames.

3:43

Scene two ends there. 6 seconds 7

3:46

frames, scene three.

3:48

8 seconds 11 frames, scene four.

3:50

12 seconds 16 frames, scene five.

3:54

Every scene is born from a pause.

3:56

Not forced, not guessed, not synced

3:59

after the fact. Built around the natural

4:01

rhythm of the voice, this is why that

4:04

stickman channel crossed 7.5 million

4:06

views with still images.

4:09

Not because of fancy tools, not because

4:11

of better graphics, because of rhythm.

4:14

And rhythm starts with the voiceover.

4:17

Get this right and everything else falls

4:20

into place. Now let's build the

4:21

voiceover. Copy your complete script.

4:25

Now visit 11labs free plan.

4:28

10,000 credits every month. This That's

4:30

all you need.

4:31

Go to voices in the left sidebar. Search

4:34

for Raunak.

4:35

You'll see it. Viral, relatable, real

4:38

voice.

4:40

Add it to your library.

4:42

Now go to text-to-speech in the left

4:43

sidebar. Make sure Raunak is selected.

4:47

Paste your complete script. Hit

4:49

generate. Download the audio. Your

4:52

voiceover is ready.

4:53

Now comes the part that separates this

4:55

workflow from every other tutorial

4:57

you've watched. Remember, voiceover

5:00

first, scene second.

5:02

But how do you know exactly where each

5:03

scene should start and end?

5:06

That's where most creators guess.

5:08

You're not going to guess. Visit

5:09

faziscribe.ai.

5:11

Link is in the description. Sign up and

5:14

log in.

5:15

Free account gives you 10 minutes of AI

5:17

credit.

5:18

Once that runs out, create another free

5:20

account and keep going. Upload your

5:22

voiceover file. You'll see two modes,

5:25

fast and accuracy. Choose accuracy.

5:28

Every frame matters here.

5:30

Click transcribe audio and watch what it

5:33

does. It doesn't just transcribe your

5:34

words. It finds every pause in your

5:37

audio.

5:38

Every single one with exact timestamps.

5:41

Pause at two seconds, pause at four

5:43

seconds, pause at seven seconds.

5:46

Now you don't need to guess where scenes

5:48

begin and end.

5:49

The voice over tells you exactly.

5:52

One more thing, you may automatically

5:54

detect the language of your audio.

5:57

Hindi, English, Urdu,

5:59

any language, this workflow works for

6:02

all of them.

6:03

Now, download the timestamp script or

6:05

simply copy it. Go back to your Claude

6:08

AI conversation,

6:10

paste the full timestamp script, hit

6:12

send.

6:13

Claude now generates a detailed text to

6:15

image prompt for every single

6:17

timestamped line.

6:18

It won't give you everything at once. It

6:21

works in parts.

6:22

When the first batch is done, type next.

6:25

Keep going until every line has a

6:27

prompt. Um once all prompts are ready,

6:30

Claude will ask if you want a text file.

6:33

Type yes.

6:34

Uh within seconds,

6:36

a complete prompt file is ready to

6:38

download. Every scene, every timestamp,

6:41

every prompt

6:43

in one file.

6:44

Now, it's time to bring these prompts to

6:46

life. Uh

6:47

copy the prompt for scene one, visit

6:50

flow.google, create a new project. If

6:53

agent mode is already on, turn it off.

6:56

Select image mode. Set aspect ratio to

6:59

16:9.

7:01

Set output per prompt to one. Select

7:04

Nano Banana 2 as your image model. This

7:07

model gives you the best results uh

7:09

without hitting generation limits. Paste

7:11

scene one prompt. Hit send. Within

7:14

seconds, scene one is ready. Uh you can

7:17

repeat this manually for every scene.

7:20

But, here's the truth.

7:22

With 100 plus scenes, that process will

7:25

take hours.

7:26

There's a faster way, a completely free

7:29

faster way.

7:30

Go to Google and search Zappy Flow

7:33

Chrome extension.

7:35

Open the first Chrome Web Store link.

7:37

You'll see the extension I personally

7:38

built for you. Completely free and free

7:41

forever. One thing, make sure the

7:44

developer name says zappywala.ai.

7:47

This protects you from fake or duplicate

7:49

versions.

7:50

Add it to Chrome, pin it, launch it.

7:53

Now, follow these steps carefully. Go to

7:56

Google Flow, open a new project, turn

7:59

agent mode off.

8:01

You'll see two messages in the extension

8:03

connected to Google Flow.

8:06

Agent mode off.

8:08

That means everything is ready. Uh now

8:10

set up the flow for bulk generation,

8:12

select image mode, set aspect ratio

8:16

16:9,

8:17

set output per prompt one, set model to

8:21

nano banana two. Now, paste all your

8:24

prompts into the prompt box. One line

8:26

break between each prompt. This is

8:28

important. You can see we have have

8:31

prompts with one line break between

8:32

them. You can simply copy all the

8:34

prompts from here or simply upload the

8:37

prompt text file you downloaded earlier.

8:40

You can see 103 prompts loaded. One

8:43

setting before you run, toggle off

8:45

include serial number in file name. You

8:48

don't need this for our workflow. Set

8:50

your download folder. Click auto save

8:53

settings and make sure the first option

8:55

is turned off.

8:56

Go back to flow, adjust the random delay

8:59

if you want or leave it as it is. Now,

9:01

hit generate. Uh and this is where the

9:03

magic happens. The extension starts

9:06

generating every single image and

9:08

downloading them one by one

9:09

automatically.

9:11

And here's the best part, you don't have

9:13

to watch it. Switch tabs, open another

9:16

Chrome profile, minimize the window. It

9:19

keeps running in the background. No more

9:21

staring at a screen waiting.

9:23

Just keep one rule, do not close the

9:25

tab, the browser, or the extension.

9:28

Everything else,

9:30

you're free. Once complete, you'll see

9:33

the result. 103 prompts executed, 102

9:37

images generated successfully. One

9:40

failed, scene 15 at the 39-second

9:42

timestamp.

9:44

Don't worry, generate that one manually

9:46

using the same process. And just like

9:49

that,

9:50

all 103 scenes are ready.

9:52

Now comes the part that ties everything

9:54

together,

9:55

editing.

9:57

Open your favorite video editor,

9:59

import the voiceover file, drag it to

10:01

the timeline.

10:03

Now, import all your scene images, drag

10:05

them to the timeline in the correct

10:07

sequence. Here's how you sync them

10:09

perfectly.

10:10

Look at the file name of the next scene.

10:12

Uh it tells you the exact timestamp

10:14

where it starts.

10:16

Next scene starts at 4 seconds.

10:18

Move your playhead to exactly 4 seconds.

10:21

Trim from the right. Uh next scene

10:23

starts at 7 seconds.

10:25

Move playhead to 7 seconds. Trim again.

10:28

Sometimes you'll need to adjust the

10:30

playhead slightly.

10:31

That's normal. Take your time here. This

10:34

is the step that creates the rhythm.

10:36

This is what made that Stick Man channel

10:38

cross 7.5 million views.

10:41

Follow this process for every single

10:43

scene. Once done, do a full preview.

10:46

Watch it from the beginning, feel the

10:48

rhythm.

10:49

If it feels right, export the video.

10:52

You'll find the full video link in the

10:54

description.

10:55

Now it's time to make it go viral. Um

10:57

head back to Claude AI. You'll see it's

10:59

already asking, uh do you want to

11:01

generate metadata?

11:03

Type yes within seconds, uh a viral

11:06

title, viral description, and viral

11:08

tags.

11:09

All generated specifically for your

11:11

video.

11:12

But we still need one more thing, a

11:13

thumbnail that stops the scroll.

11:16

Type this prompt.

11:17

Ask for five high CTR thumbnail prompts

11:21

for your video with one line break

11:23

between each.

11:24

And ask Claude to provide them in a

11:26

copyable code block.

11:28

Copy all five prompts, go to the Zappy

11:31

extension, toggle it once, paste the

11:33

prompts, set your download folder, hit

11:36

generate. Five thumbnails, one click,

11:39

done.

11:40

Now go to your YouTube channel, upload

11:42

your video, use the viral metadata

11:45

Claude generated, pick your strongest

11:47

thumbnail,

11:48

publish. You just built a complete stick

11:51

man doodle video from a blank page to a

11:54

published YouTube video

11:56

using nothing but free AI tools.

11:58

No paid subscriptions,

12:00

no design skills, no team.

12:04

Just you, this workflow, and the system

12:06

that works.

12:07

If this helped you, hit the like button

12:09

and subscribe. Small creators like me

12:12

run on your support. All master prompts

12:15

and resources are in the description and

12:17

pinned comment. And if you want a

12:19

complete money printing YouTube

12:21

workflow, you cannot afford to miss the

12:24

video above. See you inside.

Continue with YouTLDR

Analyze another video with Pro

Process a new video, search every timestamp, compare sources, and keep the result in your library.

Get Pro — $12/month30-day money-back guarantee

More transcripts

Explore other videos transcribed with YouTLDR.