The short answer is two or three processes
In almost every company most of the repetitive time goes into two or three activities, and those are the ones to take, the rest is a long tail of small things that feel annoying but weigh very little
The reason is plain, an automation pays for itself through repetition, if you do something two hundred times a month every minute you take away is multiplied by two hundred, if you do it twice a month it stays twice a month afterwards too, and the time spent explaining its rules you never get back
So the first thing to do is not to make a list of everything that annoys you, it's to work out where the volume is, because annoyance and volume are not the same thing, the case that gets on your nerves may be the one that comes up rarely, and the one that really weighs on you maybe you don't even notice anymore because you've got used to it
Anyone offering to automate everything is selling you hours
Be wary of the person who turns up and tells you the whole office can be automated, because that pitch suits the one making it, not you, the more things go into the project the longer the project lasts, and the length is what you pay for
A job done well starts from the opposite end, you look at what to leave as it is, and that's the right place for the case that turns out differently every time, or the step where someone has to decide by looking at the context, it isn't giving something up, it's the reason the rest works
There's a practical reason too, an automation has to be written, tested and then maintained over time, because the business software you use changes, a new supplier arrives, a form changes, and every automated piece is a piece somebody has to keep standing, if you build ten of them for the pleasure of saying everything is automatic, you've bought ten things to maintain and you use three
After the first process you understand the second one better
There's something you only see later, once the first process is running, the second one is almost never the one you had in mind at the start
It happens because while the first one is being written you look at the work close up, step by step, and things come out that couldn't be seen from above, a file that was being checked twice by two different people, a manual step that existed only because another step needed it and that step you no longer do, and at that point the ranking changes on its own
That's why it's better to start with one, the most repeated and the clearest, make it work, watch it working for a while, and only then choose the next one with what you've learned, instead of deciding now on a list of six things you'd rewrite differently a few months from now
If you want an idea of which one comes first, which processes to automate first has the criteria for lining them up, and work out the savings shows you what the time you spend on it now weighs over a year