Don't take anything personally.
You have many ways (data structure) to express the same data, some are efficient and some are not. They implemented like 5 different data structures for user to choose themselves. But the functions inside the same program only support a specific one. They don't even bother to make functions support all data structures that they provided. You don't even know why they make such arbitrary decision to support this DS and not the others. Of course they never mentiom anything until the problem raise error to you, then you check the document, don't know how to fix it because who knows what DS it accepts?
Then there comes a separate team solely from IBM Japan that has zero communication with the original team , make their own stuff said their fork has better performance. Then the stuff became official but the API is no where compatible. You can't drop in replace to gain performance.
The whole company is completely nonsense.
You had to go back to the main form for navigation to do anything, over and over... There were arbitrary limits on notes far far too short,1000 characters isn't long enough for a complete medical history.
I hope it has improved, but I doubt it.
It is a very rare piece of non-trivial software that isn't "bad" in some manner (even most trivial software is "bad")
It really comes down to managing expectations, enduser training, known issue mitigation, etc