General Instinct is building the deployment layer for physical intelligence, helping teams take frontier vision and language models out of the cloud and onto the edge. Robotics, drone, wearable, and industrial customers often hit a wall when a large model that runs fine on a GPU server has to operate on a battery-powered device with intermittent connectivity. General Instinct exists to compress, optimize, and serve those models on real hardware so that perception and reasoning happen on device, in real time, even when the network drops.
Its flagship product, Instinct Edge, targets accelerators like NVIDIA Jetson, mobile NPUs, and ARM SoCs, and the team has already demonstrated a 15B parameter vision language model running on an iPhone 13. A hybrid inference mode lets workflows fail over between local and cloud execution so vision pipelines keep running through connectivity gaps. The platform is aimed at builders shipping AI wearables, autonomous drones, security cameras, and factory vision systems that cannot tolerate latency, bandwidth costs, or privacy exposure from constant cloud calls.
General Instinct is a Spring 2026 Y Combinator company founded by Bill Jiao and Guanming Wang, based in San Francisco. The pair are positioning the company as critical infrastructure for the wave of physical AI products coming to market, where the bottleneck is no longer model quality but reliable on-device deployment.