Turn measured competitor breakouts into native content drafts.
The workflow compares each video with that channel's own normal, keeps relevant breakouts, separates observed evidence from hypotheses and adapts one useful pattern to your brand.
- Latest validation
- 250 videos judged
- Qualified
- 44 videos, 40 concepts
- API scan
- $0, 25 quota units
One ranked concept, shaped into nine native drafts.
The workflow researches recent YouTube videos in your niche, measures what earned unusual attention on each source channel and adapts one relevant concept to your business without copying the source's proof or distinctive wording.
What you need before the run.
Business context
Your website, active channels, known competitors, audience, growth goal, facts and approved proof.
Research tools
The official YouTube API, transcript extraction, ranking rules, clustering and brand-copy checks.
Final approval
A person confirms the facts, chooses the concept, records the flagship piece and approves every draft.
Run the recipe in six steps.
The customer supplies known competitors. The current tested source is YouTube. The workflow reads recent public videos through the official API.
Each candidate is compared with the account's previous 20 videos of a similar age and format. Thin data is labelled lower confidence.
A video must clear the current 2x channel-median bar and the minimum view floor. Livestreams and very long broadcasts are removed.
The system separates measured performance from hypotheses about the hook, angle, structure, proof, pacing and payoff.
The strongest relevant concept becomes original drafts checked against the customer's facts, voice and proof boundaries.
Every asset stays a draft. A person chooses what is true, useful and worth publishing. Nothing posts automatically.
From source evidence to an adapted draft.
This trace came from a real scan and draft pack. It shows what was observed, what was inferred and what changed during adaptation.
Nate Herk's “This New AI Makes Decisions for 18 Cents” had 17,055 views and scored 20.06x against that channel's normal in the first scan on 22 September 2026. On 23 September the ranking became stricter: each video is now compared with the channel's previous 20 videos of a similar age and format. This video is not among the 44 that clear that bar. The trace is kept to show the method, not as a current ranking.
A real, specific and surprisingly small cost becomes the whole hook. This is a hypothesis about the structure, not proof that it caused the performance.
The draft used a different tool, number, audience and wording: “I read 250 competitor videos this morning. The YouTube scan cost $0.”
No source wording or personal proof was reused. The output remained a draft waiting for Luka's approval.
Cost, evidence and current limits.
Latest validation: 23 September 2026.
| Working today | Research, ranking, clustering, evidence and hypothesis separation, source links, cost ledger and copy checks. Sixteen tests passed at the first recorded proof point. |
|---|---|
| Latest validation | 250 videos judged across five channels, 44 cleared the 2x bar and became 40 concepts. Eleven livestreams were removed. The official YouTube API scan used 25 of 10,000 daily quota units and had $0 metered cost. |
| Paid step | Transcripts cost about $0.00188 each. Two transcripts used for the latest content pack cost $0.00376 in total. |
| Current limitation | The generation step remains human-assisted. Repeat delivery time has not been measured. |
| Not claimed | A source video's performance does not prove the adapted content will perform, convert leads or create paying customers for another business. |
| Evidence status | Internally tested. Not yet pilot-tested with a paying customer. |