Meta open sourced a model that simulates how the human brain reacts to video.
We used it on one client launch and got >1.8M views.
That post made the client millions...
Most founders think virality is creative.
It isn't.
It's neurological.
Every video that crosses 1M views on this platform does the same thing to a viewer's brain. It activates specific neural patterns at specific timestamps. Spike. Sustain. Payoff. Miss any of the three and the post dies in 60 seconds.
For two years our team has been engineering for this without being able to measure it. We argued about cuts. We watched retention graphs after the fact.
Then Meta dropped TRIBE v2 and it made going viral 10x more predicatable.
It's a model trained on real neurological data brain scans from 700+ people watching videos, listening to audio, and reading content. It models how 70,000+ neural signals activate in response to what someone is experiencing on screen.
You drop in a video. It tells you, second by second, how the average human brain is most likely responding to it.
The signal that took our team years to develop intuition for, Meta just released for free.
Drop in any video, and TRIBE shows you second by second how the human brain is likely to respond.
Example below:
Here's exactly how we use it:
————
Step 1: Stop treating your hook like a sentence. Treat it like a brain scan.
Most founders write a hook by feel.
"Does this sound bold?"
"Would I scroll past it?"
"Is it punchy?"
Wrong question.
The right question: where does the brain spike, and where does it flatline?
We ran one of our client launches through TRIBE before the final cut went out. The hook tested clean — sharp activation spike in the first 1.5 seconds. But the second beat (the demo intro) had a multi-second flatline before the brain re-engaged.
That flatline was the entire problem.
In the original cut, the founder said "let me show you how it works" before showing anything. The brain disengaged for the entire setup.
We cut it. Replaced it with the demo starting cold.
Same script. Same shots. One specific deletion.
1.8M views.
Here's how to apply this to your own video:
→ Run your full edit through the model
→ Find every flat zone over ~1.5 seconds
→ Cut anything where the brain isn't activating
→ Re-test the new edit
→ Don't publish until the timeline reads as continuous activation
Most launches die in the flatlines. TRIBE shows you exactly where they are before you ship.
Step 2: Use the model to win the editor argument.
This is the part most agencies will hate.
Editors are paid to make creative decisions. Pacing, cuts, music drops. The whole job is "trust me, this feels right."
But "feels right" is the editor's brain. Not the audience's brain.
TRIBE is the audience's brain — averaged across 700+ scans.
We're not replacing editors. We're giving them a feedback loop they've never had. Instead of debating whether a cut works, we run two versions through the model and let it tell us which one activates the brain harder.
Weak: "Trust me, the longer cut breathes."
Strong: "Cut A scores a 7.4 sustained activation curve. Cut B scores 9.1. We're shipping B."
The editor isn't wrong about taste. They just aren't measuring the same thing as the algorithm.
Step 3: Optimize for the loop, not just the hook.
Every viral video has three brain moments, not one:
→ The spike (first 1.5 seconds)
→ The sustain (seconds 4 through 20)
→ The payoff (last 3 seconds)
Most launches fixate on the first one and ignore the other two.
That's why so many "open big and then die." The hook scores high. The body and ending don't carry the brain forward.
TRIBE shows all three curves at once. You can see exactly where the engagement drops, where it spikes back up, where the payoff lands.
On the $1M launch, our first edit had a textbook hook spike. But the payoff at the end was flat. The viewer's brain disengaged before they ever saw the CTA.
We rebuilt the last beats specifically to spike the payoff curve. [Matt: drop in the actual edit you made — e.g. "tightened the final benefit + hard cut to the giveaway frame."]
That payoff curve is what drives the saves and the follow conversions. Both of which feed the algorithm directly.
Step 4: The whole setup takes 10 minutes.
Here's the entire workflow:
→ Go to aidemos.atmeta.com/tribev2
→ Upload your video file
→ Wait for the activation curve to render
→ Pull every timestamp where activation drops below threshold
→ Cut or rebuild those exact seconds
→ Re-upload. Repeat until the curve runs hot through the whole timeline.
No API. No fine-tuning. No subscription. No setup tax.
The only people losing here are the ones who don't know it exists.
————
❗ I'm putting our full TRIBE editing protocol — the exact threshold settings, the cut decision tree, and the before-and-after curves from both of our launches — in a doc. FOLLOW + comment "BRAIN" and I'll DM it to you. ❗
If you have a launch coming up and want it run by the team that's now used this to crack 1M+ views and $1M+ in closed revenue, DM me.
We build the video. We brain-scan the cut. We plug you into our distribution system. We fill your calendar.
Pricing starts at $100K.
Get Shown.
