One of my teachers used to say "what is cheaper: Robert or robot?". There already are constraint based systems for structural engineering that e.g. calculate supports given structure and loads, select materials, etc. I am not too familiar with UK market, but in my country people in civil engineering tell me that significant portion of engineering project costs come not from actual engineering, but rather various permits, coordination and other bureaucracies that do not benefit from single party automating.

This makes returns on investment into new software, trainings and lost temporary productivity diminishing. Without government changing standards on engineering output, the market for flexible projects is not that large (which is a bit of a chicken and an egg problem).