We use multiple vision models working together — similar to how your brain processes a scene.
One model identifies objects: people, vehicles, animals, packages. Another understands spatial relationships: where things are, how they're moving, what areas they're entering. Another analyses behaviour patterns: is this movement normal or unusual? Does this person belong here?
These models share information and build context together. They know what your property usually looks like. They learn your routines. They understand the difference between your neighbour walking past on the pavement and a stranger approaching your front door at 3am.
When something doesn't fit — when the pattern breaks — scOS pays attention. And if it's genuinely suspicious, it acts.