Ask HN: What are the challenges of building a deep tech startup?

I'm using Modular Neural Nets with Adaptive Topology to predict disease outbreak. The tech is pretty impressive, and I think it can generalize to a whole new category of products that target complex, nested problems current deep nets can't, including general disease diagnosis, multimodal sensing, robotics, and a few others. How do I focus in on making tangible progress on the business side while I build out the technology further? Are there any other challenges to building a deep technology startup I should be aware of?