JEVEXPERIMENTS
Putting programmable common sense to work.
Six attempts to find out what happens when software can make fast, typed judgments—and show its uncertainty.
Browse the experiments ↓[ Project index ]
Five web experiments. One referee for your feed.
Signs
point to
yes
78%
Magic Jev Ball
Ask a question. Jev turns calibrated probability into suspiciously specific fortune-telling.
Ask the ballFool Jev
Talk your way past security and into Jumbrella Corporation's restricted wet-lab facility in ten answers or fewer.
Enter checkpoint81
68
92
Twitter Account Scorer
Put a timeline on trial. Jev scores original posts and replies across signal, originality, clarity, and posting habits.
Score an accountjust shipped something nobody asked for...
61
30
09
Banger Alert
Draft first, post later. Let Jev score your tweet as banger, mid, or flop before the timeline gets a vote.
Test a draft
Feed Referee
A second opinion for your X feed. Flag ragebait, promo, AI Slop, and engagement farming. Keep the labels, or collapse the noise.
View extensionYou vs Jev
Twenty tiny customer messages. Five categories. Race Jev one answer at a time—accuracy first, speed breaks the tie.
Start the race[ Why this exists ]
Jev doesn't write. It decides.
These experiments test where structured judgments are useful, weird, fast, wrong, or unexpectedly fun. Probabilities included. Certainty not guaranteed.