I.Mission Control

Mission Control

Status of every agent, every memory, every signal.

·Local · Bangkok

Gemma 4

local-first · on-device · private AI
Live readinesschecking…
Ollama
not detected
Gemma 4 model
—
7.2 → 4.3 GB
QAT on disk
12–31% faster
on M2 Air
E2B–31B
sizes on Ollama
Local-AI demos
AI Pictionary

Draw on the canvas — local multimodal Gemma 4 guesses what it is, fully on-device.

needs Gemma 4open full
Local Oracle

Ask anything. A reactive orb pulses as the private model thinks and answers.

needs Gemma 4open full
QAT Visualizer

See how quantization-aware training shrinks the model — 7.2GB to 4.3GB, same quality.

needs Gemma 4open full