Video summary

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

Main summary

Key takeaways

Technology

High-level concept of the video

The video claims you can create “stick man” / doodle-style YouTube videos (static images stitched together) using a fully free-ish, end-to-end AI workflow, covering:

  • ideation → script → voiceover
  • timestamp-aligned scene prompts → image generation
  • editing sync → viral metadata + thumbnails → upload

The core argument is that these videos succeed due to rhythm, not graphics or animation.

Key analysis / product-feature claims (technological ideas)

1. Main differentiation: “Voice-over first, scenes second.”

The creator argues most workflows generate scenes/images first and then create the voiceover, which causes a “collision” that breaks viewer rhythm.

Correct workflow:

  • Generate voiceover first
  • Build scenes around natural pauses in the audio
  • Each scene is born from a pause
  • Editing becomes timestamp-driven rather than guessed

2. Single prompt to Claude (free plan) for ideation + scripts

  • Uses the Claude AI free plan with one prompt (as stated in a pinned comment).
  • Claude returns 5 viral topic ideas.
  • After selecting one topic, Claude generates a long-form script (claimed to be ~5–15 minutes).
  • Claude also outputs a downloadable text file containing the script.

3. Free text-to-speech via 11Labs free plan

  • Uploads the script to 11Labs using the free plan.
  • Selects a voice named “Raunak”.
  • Generates the full voiceover audio, which can be downloaded.

4. Timestamp extraction from the voiceover using faziscribe.ai

  • Uploads the voiceover to faziscribe.ai.
  • Uses an “accuracy” mode to produce:
    • transcripts with exact timestamps for pauses
  • Claims it can auto-detect languages (e.g., Hindi/English/Urdu).
  • Produces a timestamp script, which is fed back into Claude.

5. Claude converts timestamped lines into image prompts (scene-by-scene)

  • Claude receives the timestamp script.
  • Generates detailed text-to-image prompts for every timestamped line.
  • Prompts are produced in batches, continued by typing “next.”
  • Claude can assemble everything into a downloadable prompt file (one prompt per scene/timestamp).

6. Bulk image generation using a Chrome extension for Google Flow

  • Uses flow.google with “agent mode” turned off.
  • Configures image generation:
    • aspect ratio: 16:9
    • output per prompt: 1
    • image model: “Nano Banana 2”
  • Uses a Zappy Flow Chrome extension to generate in bulk.
  • Prompts are supplied by:
    • pasting into the extension (prompts separated by one line break), or
    • uploading the downloaded prompt text file
  • Mentions important toggles:
    • developer name should be zappywala.ai (to avoid fake extensions)
    • “include serial number in file name” should be toggled off
    • random delay is configurable
    • auto-save settings adjusted (first option turned off)
  • Generation runs in the background; you can switch tabs/minimize, but must not close the tab/browser/extension.

7. Editing sync approach based on filename timestamps

  • In the editor, place the voiceover on the timeline.
  • Import scene images and arrange them using their timestamps embedded in filenames.
  • Method:
    • move playhead to the next scene start time
    • trim the scene image clip to match
    • iterate for each timestamp to preserve rhythm

8. Viral metadata and thumbnail generation using Claude

  • Claude generates YouTube metadata:
    • title, description, tags
  • For thumbnails:
    • requests 5 high CTR thumbnail prompt ideas
    • requires prompts in a copyable code block
  • Then uses the same Zappy extension workflow to generate 5 thumbnails with one click.
  • User selects the best thumbnail and publishes.

Practical workflow checklist (tutorial-style steps)

  1. Claude ideation + script

    • Open Claude AI free plan
    • Paste the single “master prompt”
    • Get 5 viral topics
    • Select one → generate script (downloadable text)

  2. Voiceover (11Labs)

    • Use 11Labs free plan
    • Select voice “Raunak”
    • Generate and download audio

  3. Transcription + pause timestamps (faziscribe.ai)

    • Upload voiceover to faziscribe.ai
    • Transcribe with exact pause timestamps
    • Copy/download the timestamp script

  4. Image prompts (Claude)

    • Paste timestamp script into Claude
    • Generate text-to-image prompts per timestamp
    • Collect/download prompt file (batch with “next”)

  5. Bulk image generation (Chrome extension + flow.google)

    • Use Zappy Flow Chrome extension + flow.google
    • Settings:
      • agent mode off
      • image model: Nano Banana 2
      • aspect ratio: 16:9
    • Bulk-generate all scene images automatically

  6. Edit and sync

    • Place voiceover on timeline
    • Add scene images in order
    • Trim each image to its timestamp (from filename)
    • Preview and export

  7. Metadata + thumbnails

    • Use Claude-generated metadata (title/description/tags)
    • Use Claude thumbnail prompt ideas → generate 5 thumbnails
    • Pick one → upload

Notable “supporting evidence” used

  • The video opens with an example “stick man” channel:
    • claims include:
      • 7.5 million views on one video
      • 14 videos
      • 137,000 subscribers
      • 14.5 million total views
      • first upload 2 months ago
  • It emphasizes that success comes from rhythm created by voiceover pauses rather than visual complexity.

Main speakers / sources

  • Speaker: the video creator (narrator) presenting and claiming they personally built the workflow.
  • Tools/Services referenced:
    • Claude AI (Anthropic)
    • 11Labs (text-to-speech)
    • faziscribe.ai (transcription with pause timestamps)
    • flow.google (image generation workflow)
    • Zappy Flow Chrome extension (automation for Google Flow)
    • YouTube (metadata + thumbnail + upload workflow)
  • Named voice: Raunak (used in 11Labs voice selection).

Original video