Video summary
ВИРУСНАЯ ИИ НИША Ютуб: $45,000 на Психологии За 8 Минут (без воды)
Main summary
Key takeaways
Business / Growth Takeaways (YouTube Strategy for the “AI Psychology” Niche)
Niche selection rationale (demand + fit)
- Psychology pain points: focus on widely searched and highly consumed topics like anxiety, toxic relationships, and insecurity.
- Low barrier to entry: videos can be created without camera/face/voice, enabling faster iteration and scaling.
Distribution & algorithm mechanics
- Audience retention drives growth: after a certain threshold, videos see a sharp jump in views per upload.
- Viral sharing behavior: viewers tend to share these videos with each other, improving algorithmic distribution.
Monetization economics (execution-focused)
- RPM range for psychology niche: $8–$15 per 1,000 views.
- Revenue example math:
- ~3.2M views in ~1 month → ~$45,000 revenue (using the cited RPM figure)
- Scaling approach: revenue is attributed to systemic output, despite only ~60 videos.
Metrics / KPIs Mentioned
- 240,000 subscribers gained in 3 months
- One video: up to 3.2M views in 1 month
- Channel income claim: “$42,000/month (VQ)”
- RPM (psychology niche): $8–$15 per 1,000 views
- Example video revenue: ~$45,000 from 3M+ views
- Video count: ~60 videos (used to argue scaling effectiveness)
End-to-End “Playbook” / Process
1) Build the “system prompt”
- Spend hours creating a single reusable high-quality prompt for generating:
- the script
- the asset prompts
- Copy the finalized prompt into Claude.
- Expected outputs include:
- readiness confirmation
- script
- generation prompts for images/video
2) Competitive inspiration capture
- Take screenshots of the latest videos from another channel.
- Feed the screenshot into the LLM to generate a unique character for your video.
3) Character creation (Flow / image generation)
- Use the LLM’s character prompt in Flow Labs, with specific settings.
- Critical step: save the character and keep it consistent across frames.
4) Idea expansion + storyline
- Generate ~20 ideas, then select the best one (example given: choice #9).
- Generate a full video script in the target language (e.g., Russian/English).
5) Voiceover generation (Telegram bot)
- Copy the script into a voiceover bot.
- Save the generated voiceover while visuals/assets are rendered.
6) Scene-by-scene asset generation
For each scene/frame:
- Use Flow to generate video animations and images.
- Critical technical fix: drag/drop/position the character appropriately before generating each asset.
- (Described as a common mistake that causes character inconsistency.)
7) Video editing (CapCut)
- Import scenes into a timeline.
- Add the voiceover track.
- Ensure:
- audio sync
- audio clarity
- correct layering of effects
Concrete “Common Mistakes” (and Fixes)
Character inconsistency across frames
- Problem: the character changes or won’t remain consistent frame-to-frame.
- Fix: save and reuse the same character reference correctly when generating each frame/scene.
- Warning: incorrect handling can cause the “character doesn’t save” issue.
Audio/effects overlap
- Problem: track sound overlaps and interferes with voice.
- Fix in CapCut: disable or attenuate unintended track sound:
- set to -∞ / 0
- or around -28 dB
- for every frame/track to prevent voice interference.
Actionable Recommendations (Implied)
- Create and refine a reusable prompt that generates:
- script + scene plan
- prompts for visuals/assets
- voiceover-ready text
- Optimize for speed-to-upload:
- generate voiceover while visuals render
- then assemble scenes quickly
- Prioritize retention and shareability:
- choose psychology topics that drive watch time and interpersonal sharing
- Maintain production consistency:
- keep the same character across all frames
- avoid editing mistakes that break consistency in CapCut
Presenters / Sources
- Presenter: referred to as “I” (name not explicitly stated in subtitles)
- Tools / sources mentioned:
- YouTube
- Claude (LLM)
- Flow Labs (image/video generation)
- Telegram bot (voiceover)
- CapCut
- VQ (used for the “$42,000/month” reference)
- “Neroshi / Nerosel” (LLM/agent referenced in the in-character generation workflow prompt)