Automating a broken process just makes it break faster. Here's a simple framework to decide what to fix first.
Automating a broken process just makes it break faster — at scale. The order matters: optimize first, then automate the thing that's actually worth running.
A simple rule
If a process is messy, slow, or full of exceptions, simplify it by hand until it's boring and predictable. Once it's boring, it's safe to automate. Boring is the goal.