Automation is not a new thing, in fact it's been around for the most part of human history. We like to make things easier for ourselves in general. The first true case of 'automation' (that which uses some kind of control) was probably the water clock (200's BCE), and later the thermostat (1620).
However, more recently work has been focused on automating that which was traditionally thought that only a human could do on a more abstract level. A simple example of this is filling in taxes or diagnosing health problems. Now multiple automated systems exist for these that reduce the human time taken to complete these tasks.
As computer systems get more and more advanced, people are able to create more and more complex forms of automation. This computerised automation was once as simple as calculation of basic arithmetic. Automation is often linked to industrial manufacture and the first electronically automated factories were implemented around the 1950's. Now some factories run completely autonomously (see: lights out manufacturing).
Hence, my spurious prediction of how the automation process will take form in the short-term.
This make sense looking at it, we're in the part where we're developing more and more complex systems to do this, soon I think we'll see automation as a service. Companies will offer bespoke automation at a massive level, digital systems will be revolutionised by this as different building blocks can be put together, much like the industrial automation where modular systems are developed and fitted together by the user, this will exist in the computerised automation of processing data. And, to and extent it already it, optical character recognition libraries already exist and are implemented on a wide scale.
I think it is just a matter of a few years before we see a company that offers this as a service that is easily deliverable and configurable. Whether it will work well at first or not is to be seen but I definitely feel it's at least the short-term future of data automation.