312 AI reels, scored against their creator's normal.
I had my AI go through 312 reels from 14 AI creators. Half had a comment keyword. Then it searched every app for posts that beat their creator's normal, and the biggest of those mostly didn't. Here's the whole sheet.
Give this to your AIClaude Code, Codex, Cursor. No signup needed.
paste this to your agent
Read https://jay-burnett-packs.vercel.app/count/agent.md and set this up for me.
It reads agent.md: install, test and undo steps. Want it by hand instead? Your email below opens the full page.
312reels
14creators
155had a keyword CTA
6things outliers shared
You're in, the files are below.
The full sheet
Every reel: creator, link, date, plays, that creator's median, the multiple, length, keyword CTA yes or no, and the first line.
Posts that did at least 3x their creator's normal, on Instagram, X, TikTok and YouTube Shorts.
1. The AI is a character or a contestant
Two AIs in a build battle, a desktop pet you talk to, characters acting out the workflow.
2. The creator's own stakes over time
Their own app, their own money, their own house, shown day after day.
3. Non-tech people on camera
A barber, a security guard, people on the street. The niche almost never shows one.
4. Silent visual proof
The finished thing is the hook. A 3D rebuild of a news event, a game shown with no talking.
5. Length isn't the lever
Winners ran from 12 seconds to 20 minutes.
6. None relies on a keyword gate
That's the cross-app outliers. Where they asked for anything, it was a vote or "follow along". Inside the 312 IG reels, the top 4 by multiple did use a keyword, so it's mixed.
Credit: @siyabuilt (AI build battle in Minecraft, 7.26M plays on X, 513x his median), Bilawal Sidhu (3D rebuild of a news event, 6.7M on YouTube Shorts, 327x), @aidendjohnston (daily app counter, 468K on Instagram, 29.7x).
Creator medians
Instagram, X and YouTube Shorts accounts in the study. Full tables are in the file below.