Speaker A
It's 2:00 a.m. at night. You can't sleep. You're scrolling YouTube, thumb moving automatically, not really looking for anything. Um, and then one thumbnail stops you cold. A stick man, a simple hand-drawn stick man, 7.5 million views.
Learn how to create viral YouTube videos using a free AI workflow with Claude AI and no paid tools, generating scripts, voice-overs, and images efficiently.
Key Takeaways
- Voice-over first, scene second is critical for creating engaging video rhythm.
- Free AI tools like Claude AI, 11labs, and Faziscribe enable fully free video production workflows.
- Automating image generation with the Zappy Flow Chrome extension saves hours of manual work.
- Natural pauses in voice-over audio define scene transitions for better viewer retention.
- No fancy graphics or paid subscriptions are needed to achieve millions of views.
Summary
- The video reveals a free, no-paid-subscription workflow to create viral YouTube videos using AI tools.
- It emphasizes starting with voice-over before generating scenes to create natural rhythm and engagement.
- Claude AI free plan is used to generate multiple viral video script ideas and complete long-form scripts.
- 11labs free plan provides a realistic voice-over using a recommended voice named Raunak.
- Faziscribe.ai transcribes voice-over audio with exact timestamps to identify natural scene breaks.
- Claude AI then converts timestamped script lines into detailed text-to-image prompts.
- Google Flow and the Nano Banana 2 model generate images from prompts, with bulk generation automated by the free Zappy Flow Chrome extension.
- The workflow avoids common pitfalls by syncing scenes to voice-over pauses rather than forcing synchronization.
- The system requires no design skills, paid tools, or teams, making it accessible to solo creators.
- The video also guides on generating high CTR thumbnail prompts and metadata using Claude AI.
Chapters
- 00:00Introduction and Viral Stickman Channel Inspiration
- 00:50Motivation to Create Viral Videos
- 01:17Overview of Free AI Workflow
- 01:55Generating Video Scripts with Claude AI Free Plan
- 02:24Common Workflow Mistakes and Correct Approach
- 03:16Building Voice-Over Using 11labs Free Plan
- 05:16Using Faziscribe for Timestamped Scene Breaks
- 06:13Generating Text-to-Image Prompts with Claude AI
- 06:45Creating Images with Google Flow and Nano Banana 2 Model
- 07:22Automating Bulk Image Generation with Zappy Flow Extension
Full Transcript — Download SRT & Markdown
Speaker A
Um, you tap it. No fancy animation, no 3D characters, no cinematic visuals, not even video clips, just still images of stick figures one after another, uh, stitched together into a full video. Um, that's it. Static images, stick figures with 7.5 million
Speaker A
views. You go to the channel, 14 videos, 137,000 subscribers, 14.5 million total views, uh, with first video published just 2 months ago.
Speaker A
You sit up. One thought hits you. If a stick man channel can do this in 2 months, why can't I?
Speaker A
You open your laptop, you try to build the workflow, but you hit a wall at the very first step. You don't know how to write a script that people actually want to watch. And every AI tool that can
Speaker A
help requires a paid subscription. You close the laptop. That stick man channel just crossed another million views while you're still at zero.
Speaker A
I know that feeling exactly, 'cause I was sitting in the same place, uh, until I built a complete free workflow that creates these videos from script to final upload. No paid tools, no design skills, just a system that works. But
Speaker A
before we dive in, there's one thing you need. A single prompt, not five tools, [music] not a paid subscription, not a complicated setup, just one prompt.
Speaker A
I spent weeks engineering and optimizing this specifically for this workflow. Uh, it's in the pinned comment below. Copy it and now open Claude AI free plan. That's all you need.
Speaker A
You can see I'm on the free plan right now. Paste the prompt and watch what happens next.
Speaker A
The prompt loads. And Claude doesn't just give you one idea, it gives you five. Five viral topic ideas.
Speaker A
Each one engineered to explode if you follow this workflow completely. Pick your number, hit send.
Speaker A
What comes back will surprise you. Not just a script, a complete long-form video script like anywhere from 5 to 15 minutes ready to use.
Speaker A
And it doesn't stop there. Claude automatically creates a text file you can download instantly. Here's the download file. Now, before we move to the next step, I need to stop you.
Speaker A
Because what I'm about to tell you next is the single most important thing in this entire video.
Speaker A
Most creators follow this workflow: write script, create text-to-image prompt for each scene, generate scenes, create voice-over, then try to sync everything together.
Speaker A
Sounds logical, right? It's completely wrong. And if you've been following the same workflow, this is exactly why your videos aren't exploding. Here's the problem.
Speaker A
When you generate scenes first and create the voice-over second, you're forcing two things together that were never built for each other.
Speaker A
You're not creating rhythm, you're creating a collision, and viewers feel that collision even if they can't explain why they clicked away.
Speaker A
The correct workflow is the opposite. Voice-over first, scene second, always. Here's why. Look at this voice-over. You can hear the natural pauses between sentences.
Speaker A
That pause at 1 second 12 frames. That's where scene one ends. The next pause at 3 seconds 18 frames.
Speaker A
Scene two ends there. 6 seconds 7 frames, scene three. 8 seconds 11 frames, scene four.
Speaker A
12 seconds 16 frames, scene five. Every scene is born from a pause. Not forced, not guessed, not synced after the fact. Built around the natural rhythm of the voice, this is why that stickman channel crossed 7.5 million views with still images.
Speaker A
Not because of fancy tools, not because of better graphics, because of rhythm. And rhythm starts with the voice-over.
Speaker A
Get this right and everything else falls into place. Now let's build the voice-over. Copy your complete script.
Speaker A
Now visit 11labs free plan. 10,000 credits every month. That's all you need. Go to voices in the left sidebar. Search for Raunak.
Speaker A
You'll see it. Viral, relatable, real voice. Add it to your library. Now go to text-to-speech in the left sidebar. Make sure Raunak is selected.
Speaker A
Paste your complete script. Hit generate. Download the audio. Your voice-over is ready. Now comes the part that separates this workflow from every other tutorial you've watched. Remember, voice-over first, scene second.
Speaker A
But how do you know exactly where each scene should start and end? That's where most creators guess.
Speaker A
You're not going to guess. Visit faziscribe.ai. Link is in the description. Sign up and log in.
Speaker A
Free account gives you 10 minutes of AI credit. Once that runs out, create another free account and keep going. Upload your voice-over file. You'll see two modes, fast and accuracy. Choose accuracy.
Speaker A
Every frame matters here. Click transcribe audio and watch what it does. It doesn't just transcribe your words. It finds every pause in your audio.
Speaker A
Every single one with exact timestamps. Pause at two seconds, pause at four seconds, pause at seven seconds.
Speaker A
Now you don't need to guess where scenes begin and end. The voice-over tells you exactly.
Speaker A
One more thing, you may automatically detect the language of your audio. Hindi, English, Urdu, any language, this workflow works for all of them.
Speaker A
Now, download the timestamp script or simply copy it. Go back to your Claude AI conversation, paste the full timestamp script, hit send.
Speaker A
Claude now generates a detailed text-to-image prompt for every single timestamped line. It won't give you everything at once. It works in parts.
Speaker A
When the first batch is done, type next. Keep going until every line has a prompt. Um, once all prompts are ready, Claude will ask if you want a text file.
Speaker A
Type yes. Uh, within seconds, a complete prompt file is ready to download. Every scene, every timestamp, every prompt in one file.
Speaker A
Now, it's time to bring these prompts to life. Uh, copy the prompt for scene one, visit flow.google, create a new project. If agent mode is already on, turn it off.
Speaker A
Select image mode. Set aspect ratio to 16:9. Set output per prompt to one. Select Nano Banana 2 as your image model. This model gives you the best results, uh, without hitting generation limits. Paste scene one prompt. Hit send. Within
Speaker A
seconds, scene one is ready. Uh, you can repeat this manually for every scene. But, here's the truth.
Speaker A
With 100 plus scenes, that process will take hours. There's a faster way, a completely free faster way.
Speaker A
Go to Google and search Zappy Flow Chrome extension. Open the first Chrome Web Store link.
Speaker A
You'll see the extension I personally built for you. Completely free and free forever. One thing, make sure the developer name says zappywala.ai.
Speaker A
This protects you from fake or duplicate versions. Add it to Chrome, pin it, launch it.
Speaker A
Now, follow these steps carefully. Go to Google Flow, open a new project, turn agent mode off.
Speaker A
You'll see two messages in the extension connected to Google Flow. Agent mode off. That means everything is ready. Uh, now set up the flow for bulk generation, select image mode, set aspect ratio 16:9, set output per prompt one, set model to
Speaker A
Nano Banana 2. Now, paste all your prompts into the prompt box. One line break between each prompt. This is important. You can see we have prompts with one line break between them. You can simply copy all the
Speaker A
prompts from here or simply upload the prompt text file you downloaded earlier. You can see 103 prompts loaded. One setting before you run, toggle off include serial number in file name. You don't need this for our workflow. Set
Speaker A
your download folder. Click auto save settings and make sure the first option is turned off.
Speaker A
Go back to flow, adjust the random delay if you want or leave it as it is. Now, hit generate. Uh, and this is where the magic happens. The extension starts generating every single image and downloading them one by one
Speaker A
automatically. And here's the best part, you don't have to watch it. Switch tabs, open another Chrome profile, minimize the window. It keeps running in the background. No more staring at a screen waiting.
Speaker A
Just keep one rule, do not close the tab, the browser.
Speaker A
Don't worry, generate that one manually using the same process. And just like that, all 103 scenes are ready.
Speaker A
Now comes the part that ties everything together, editing. Open your favorite video editor, import the voiceover file, drag it to the timeline.
Speaker A
Now, import all your scene images, drag them to the timeline in the correct sequence. Here's how you sync them perfectly.
Speaker A
Look at the file name of the next scene. Uh it tells you the exact timestamp where it starts.
Speaker A
Next scene starts at 4 seconds. Move your playhead to exactly 4 seconds. Trim from the right. Uh next scene starts at 7 seconds.
Speaker A
Move playhead to 7 seconds. Trim again. Sometimes you'll need to adjust the playhead slightly.
Speaker A
That's normal. Take your time here. This is the step that creates the rhythm. This is what made that Stick Man channel cross 7.5 million views.
Speaker A
Follow this process for every single scene. Once done, do a full preview. Watch it from the beginning, feel the rhythm.
Speaker A
If it feels right, export the video. You'll find the full video link in the description.
Speaker A
Now it's time to make it go viral. Um head back to Claude AI. You'll see it's already asking, uh do you want to generate metadata?
Speaker A
Type yes within seconds, uh a viral title, viral description, and viral tags. All generated specifically for your video.
Speaker A
But we still need one more thing, a thumbnail that stops the scroll. Type this prompt.
Speaker A
Ask for five high CTR thumbnail prompts for your video with one line break between each.
Speaker A
And ask Claude to provide them in a copyable code block. Copy all five prompts, go to the Zappy extension, toggle it once, paste the prompts, set your download folder, hit generate. Five thumbnails, one click, done.
Speaker A
Now go to your YouTube channel, upload your video, use the viral metadata Claude generated, pick your strongest thumbnail, publish. You just built a complete stick man doodle video from a blank page to a published YouTube video using nothing but free AI tools.
Speaker A
No paid subscriptions, no design skills, no team. Just you, this workflow, and the system that works.
Speaker A
If this helped you, hit the like button and subscribe. Small creators like me run on your support. All master prompts and resources are in the description and pinned comment. And if you want a complete money printing YouTube workflow, you cannot afford to miss the
Speaker A
video above. See you inside.
Topics:Claude AIfree AI toolsYouTube video creationvoice-over firsttext-to-image promptsZappy Flow extensionviral video scriptAI video workflowstickman channelcontent creation automation
