The method · 3 Oct 2026
The first frame decides: 43.6% stayed versus 12.9%
Across two batches of our own Shorts, the opening picture mattered more than the hook's wording. Objects you understand instantly beat screens and ordinary scenes.
AI faceless video generator for YouTube Shorts
Free to set up. No account needed until the end.

Hi, I'm the Faceless Scientist. I make faceless Shorts for your YouTube channel, then test every batch and keep what people watch longer.
Start your channelWhy experiments
Most AI tools pick one template and post it forever. Viewers learn the look and swipe past.
Since July 2025, YouTube can stop paying channels whose videos look mass-produced. In January 2026 it shut down some of the biggest ones.
Every batch tries something new, and only proven winners stay. Your videos get more original and more watchable over time.
Does the first frame decide who keeps watching?
Share of viewers who stayed instead of swiping away
Objects beat screens. One video each, so it's a hint, not a result. Testing it properly next.
How a batch works
Each stage has its own AI agent. Review every step, or switch on autopilot and just check the results.
Reads the last batch's numbers, decides which side won, and picks the next change to test.
Finds true stories for your topic and writes a script for every video, split between the two sides of the test.
Builds each Short: voice, pictures, word-by-word captions and motion, all in your channel style.
Schedules the batch to YouTube, alternating A and B so the time of day can’t skew the result.
What it tests
Each batch tests one of these. The winner becomes part of your channel's style, with a note of the batch that proved it.
The rules it follows
Everything else stays the same, so the result means something.
One video proves nothing. In our first batch, each side had one hit and one dud.
Since March 2025 a Short counts a view the moment it starts. What matters is how many stayed.
One runaway video shouldn’t decide a test.
Keep the current style, test the next thing, come back later.
Lab log
Every batch we post goes in the lab log: what we tested, both sides, the real numbers and what we changed. Tests that lost stay in too. Follow along as the channel grows from zero.
Open the lab logFrom the blog
The method · 3 Oct 2026
Across two batches of our own Shorts, the opening picture mattered more than the hook's wording. Objects you understand instantly beat screens and ordinary scenes.
The method · 3 Oct 2026
In our first test batch, each side had one Short that took off and one that flopped. Here's how many videos a fair test needs, and why.
Money and rules · 3 Oct 2026
What YouTube means by mass-produced or repetitive content, which faceless channels it hits, and how to stay monetised.