The fact that they are targeting central stations who monitor video for residential markets is a misstep. The residential market is a self-monitoring DIY market.
It is great they have a plan but the reality of providing high quality (human) video annotations to train an algorithm is a massive undertaking in terms of time and accuracy.
Putting "another box" to do the processing on the edge is similar strategy that others in the same space have taken.... however it is more likely that only commercial customers would tolerate the additional cost of this implementation.
I wish them lots of luck but sounds like nFlux strategy is still in flux.
I had a similar confusion when I first saw their targeting.
Generally speaking, I think tech driven new entrants have their best shot at the enterprise where more advanced features are more highly desired. In the consumer / residential market, generally, it is less about high technology and more about packaging (think Dropcam and Ring). Indeed, often the tech can be overkill for residential.
I don't know enough about nFlux but if they can do what is described here, I think they would find good interest from the enterprise side.