It sees what happens next.
Intelligence starts in the world, not in words. The Eye turns camera and sound into a compact picture of where things are and how they move. It learns on the device how that picture changes, then imagines the next seconds and the result of an action before the machine takes it. It never draws pixels, and nothing it sees leaves the Pi.
Sees in space, not words
Each moment becomes 104 numbers: brightness and movement in an 8 × 6 grid, plus eight bands of sound. Predictions happen there, a dimension a sentence can’t hold.
Learns on the device
There’s no training run and no cloud. Every frame updates the predictor. “Prediction skill” shows how much better it does than guessing that nothing changes.
Imagines before acting
Each time a skill runs, the Eye notes what changed. Ask it to imagine, and it compares the future with and without that action. The local AI uses this in chat to explain and decide.