LLMs are increasingly applied to unstructured data - this has been one of the key moments for AI adoption ;
Enterprise AI applications require explainable, robust and secure models powering them - this is one of the biggest blockers at this point in time ;
The ability to take LLMs and generate structured data out of them to feed into more simple, explainable models is vastly unexplored ;
The achieved data interoperability is a door-opener for new ways of training models across organizational and geographical boundaries - as well as leveraging techniques such as federation.