It's all about pretty short term returns of course.
You can build a web interface separately (external contract ofc) and that will work in parallell, but aside from very basic services there are tons of edge cases, bad documentation, business customers with systems built with yours as a dependency. It's a little like any bank old enough still having paper file storage and the planned end of life is the literal death of all paper holding customers.
So the old system sticks around and as long as it does noone wants to buy a new set of computers to run in parallell with the old ones.
There are companies who try hard and end up sinking hundreds of millions of dollars into replacing old systems that only cost a server blade, ten devs and five people in operations to run.
With no clear benefit visible to the shareholders far from all of them have even started that move.
If I can book it through the web (hence, they have a web frontend) they could have though of something better in the last 10 years
Or, you know. Make shift-tab work. And use a faster terminal emulator