I classify them as deeptech because they are doing embedded...

I classify them as deeptech because they are doing embedded multimodal ML to a high enough standard to build reliable products. That is extremely hard to do, and the longer they do it the greater their competitive advantage (they have far better data by now than the average kaggle competition would provide). This is overlooking the fact that reliably detecting drones at all is hard.But you don’t have to take my word for it; take Australia’s largest VC. Blackbird just ran the Foundry deeptech accelerator a couple of months ago, and one of the other participants was doing embedded machine learning. They considered that deeptech. I wasn’t even doing FPGAs, my company (also participating) was merely doing a novel approach to machine learning, and it was still considered deeptech. Droneshield easily qualifies. Their ability to detect and track is what matters. The jamming is just one application of that.As an aside, regarding your point about autonomous drones and jamming; RF jamming is going to be needed everywhere to degrade a swarm’s ability to coordinate. Jamming will be combined with other sorts of things (case in point, Epirus and Droneshield working together).Anyway, also worth considering is that embedded ML is a force multiplier for everything else. The reason they can get into electronic warfare and SDRs so easily is because they can already do everything involved.