Modern measures of risk

What is everyone using nowadays? For example, stuff like VAR is considered pretty "quaint", market Beta has been often derided as being questionable as a measure and whatnot, so I was wondering what the guys on the ground actually care about.

Hazarding a guess: We've had massive expansion in computational power and access to data, so I was thinking today's models would be far more granular. Lots of new modeling tricks with machine learning techniques such as DNNs and GANs. We can probably analyze network-based risk models much more effectively now as well.

I assume a lot of this stuff is proprietary, but if someone who actually does this stuff can chip in, that would be very nice.