Basically the mainframes were almost incapable of any data exchange that was meaningful in the 21st century. Instead of IBM doing something same the market just built software to scrape bills off the mainframes printer output, which you could intercept in electronic form. But then every accounts bill files were different. And sometimes the mainframes moved data around on the page, and so on and so forth.
That gives me flashbacks. I worked on a payment system that had to pull data from third-party service providers that only made it available in the form of downloadable spreadsheets from godawful slow and unreliable JavaScript-heavy web portals. Just getting to the point of downloading the spreadsheets was painful enough, and then the spreadsheets themselves were inconsistent and would sometimes flap back and forth between two different formats for a few weeks as the service provider repeatedly deployed and then rolled back new versions of their software.
