About PrePost

PrePost predicts how an average viewer's brain responds to a short video, second by second, and explains it in plain English. It runs Meta's TRIBE v2 brain-encoding model on the video file alone, so a creator or the person approving their draft can see what lands before it posts.

We're a six-person team that started in WSU's ENTRP 490 in fall 2026.

Our mission

Turn “post and pray” into “look before you leap,” and never claim more than we can show.

Short videos are won or lost in seconds, and feedback on a draft is usually gut feel. We give creators and reviewers a second-by-second read they can point at. We also publish our own evidence, including a first pilot that didn't prove the link to engagement (r = 0.25, p = 0.24, 24 videos), and we call PrePost a diagnostic until a pre-registered test shows it predicts more.

Honest

We state the evidence, including the nulls. Every read is labeled a prediction, because that is what it is.

Curious

Every clip is a question worth asking: which second holds a viewer, and why?

Plain-spoken

Neuroscience in the words a creator uses on set. If a term needs a gloss, it gets one.

Rigorous

Pre-registered tests, published methods and real model output only. No made-up demo numbers.

The team

The six people building PrePost.

See what we're building