In a low-slung office on Great Oaks Boulevard in San Jose, a team of fewer than sixty people writes software that has almost certainly watched you. Not the company itself - you have probably never heard its name - but the analytics engine it licenses, quietly embedded in a doorbell camera, an ATM lobby, a parking gate, or a city intersection. IntelliVision does not sell you a product. It sells the intelligence that goes inside somebody else's.
Founded in 2002 by Vaidhi Nathan, IntelliVision was building artificial intelligence for cameras years before "deep learning" became a conference keyword. Its core idea has stayed constant: a camera should not just record pixels, it should recognize what those pixels mean - a face, a license plate, a person climbing a fence, the sound of breaking glass. The company turns that recognition into software that runs three ways: on the tiny chip inside the camera (the edge), on a server, or in the cloud.
That flexibility is the point. A smart-home OEM wants analytics small enough to fit on a battery-powered doorbell. A bank wants them hardened for an ATM. A city wants them scaled across thousands of traffic cameras. IntelliVision built one analytics stack that can be tuned for all three, then licensed it to the manufacturers, chipmakers and integrators who ship the hardware.
