Sunny Sethi does not fit the stereotype of a firefighting revolutionary. A soft-spoken PhD in polymer science, he spent years moving through nanotechnology, solar, semiconductors, and automotive manufacturing before turning his attention to one of the most stubbornly unchanged tools in public safety: the fire hose nozzle.

From his base in Hayward, California, Sethi’s company HEN Technologies claims to do what legacy manufacturers have not: put out fires up to three times faster while using roughly a third of the water. The breakthrough came from computational fluid dynamics work funded in part by the National Science Foundation, modeling how water droplets behave in high heat and high wind. HEN’s nozzles precisely control droplet size and velocity, keeping streams coherent instead of letting them atomize and drift away.

But the nozzle, as Sethi likes to say, is just “the muscle on the ground.” Around it, HEN has built a full stack of hardware: monitors, valves, overhead sprinklers, pressure and flow-control devices, all embedded with custom circuit boards, sensors, and edge computing chips. Twenty-three different board designs turn what used to be dumb metal into a network of smart endpoints.

Those endpoints feed a cloud platform that tracks, in real time, how water moves through a fire scene: which hydrant is in use, how much water each engine is drawing, what pressure is available, how wind and weather are shifting. The system can warn incident commanders that a truck is about to run dry or that a wind change could trap crews if they do not reposition.

For fire departments, this is operational gold. It addresses a chronic blind spot in major incidents, where mismanaged water supply has contributed to disasters from Oakland to the Palisades. For Sethi, it is something else as well: a unique, high-fidelity dataset on how fire and water interact in the real world.

Every deployment generates multimodal data on fluid dynamics under extreme conditions, exactly the kind of information AI researchers need to build “world models” that understand physical reality rather than just pixels and text. Simulations can approximate this, but they cannot fully replicate the chaos of an actual wildfire or industrial blaze.

HEN is already selling to thousands of departments and major federal and military customers, and investors have taken notice, backing the company with tens of millions in venture funding. Sethi is still talking about better nozzles and safer evacuations. Yet behind the hose streams and hydrants, he is quietly assembling something far more valuable: a proprietary fire-and-water dataset that could power the next generation of physics-aware AI.