YOUTUBE CHANNEL ANALYSIS
80.4K subscribers · 100 videos · typical long-form upload earns 6.4K views, from a sample of 96.
BREAKOUT VIDEOS
Each upload is measured against what this channel was typically doing around the time it went up, with the head start an older video has taken out. Shorts and livestreams are excluded.
Provocative theoretical or structural critiques outperforming neutral news (e.g., 'The AI Scaling Problem' at 66.8x vs 'RL Foundation Models Are Coming!' at 0.9x)
Clear technical explanations framed around specific fundamental algorithms (e.g., 'Proximal Policy Optimization Explained' at 34.7x and 'Inverse Reinforcement Learning Explained' at 6.9x)
Direct technical feasibility tests targeting specific AI tools or papers (e.g., 'Can OpenAI Codex Create AI?' at 13.5x)
Tight focus on core limitations or bottlenecks rather than speculative roadmaps (e.g., 'This is What Limits Current LLMs' at 5.0x vs '12 Steps to AGI' at 1.0x)
Your viewers tune in for grounded engineering depth rather than broad tech commentary or speculative timelines. The massive success of The AI Scaling Problem at 66.8x median and This is What Limits Current LLMs at 5.0x shows that your audience wants sharp, technical critiques of industry assumptions. Contrast this with speculative or roadmap videos like 12 Steps to AGI and AGI is NOT coming soon, both of which hovered around 1.0x median. Viewers care deeply when you dissect why a specific technical approach hits a ceiling, not when you discuss distant philosophical outcomes.
Specific named algorithms and practical code tests consistently drive strong baseline traffic over long horizons. Proximal Policy Optimization Explained achieved 34.7x median over time, and Can OpenAI Codex Create AI? hit 13.5x, whereas general news updates like RL Foundation Models Are Coming! sitting at 0.9x failed to generate long-term search or core interest. You should double down on precise technical breakdowns of breaking foundational architectures and hands-on capability experiments, while completely abandoning vague future forecasts and broad paper summaries.
ORIGINAL IDEAS FROM THESE PATTERNS
Applies the exact framing pattern of the 66.8x breakout 'The AI Scaling Problem' to analyze technical limits in context length vs attention quadratic cost.
Mirrors the channel's highest-performing title structure while focusing on an urgent technical bottleneck.
Applies the precise naming structure of 'Proximal Policy Optimization Explained' (34.7x) to DPO, the modern mathematical alternative to RLHF.
Directly targets search intent for a major RL replacement using a proven video format.
Adapts the high-performing bottleneck framework from 'This is What Limits Current LLMs' (5.0x) to DiT architectures like Sora and Stable Diffusion 3.
Strong technical hook targeting high interest in generative video/image backbones.
Recreates the direct capability experiment format seen in 'Can OpenAI Codex Create AI?' (13.5x) using concrete benchmarks.
Concrete subject with a clear technical benchmark payoff, though slightly lower search longevity.
Extends the algorithm breakdown format of 'Inverse Reinforcement Learning Explained' (6.9x) to DeepSeek's novel GRPO algorithm.
Highly relevant algorithm, but niche mathematical topic limits initial click-through breadth.
Focuses on hardware memory bandwidth bottlenecks in inference, matching the architectural limitation pattern of the channel's top videos.
Extremely detailed technical subject, but the title lacks a proven viral hook structure.
Cutline finds the videos that beat your baseline, explains the pattern behind them, and turns it into an original script plus posts for LinkedIn, X and Instagram.
Analyse your channel free ↗Based on public YouTube data for the 96 most recent long-form uploads. Updated 2026-09-03. Cutline is not affiliated with YouTube or Edan Meyer.