Automation is appealing precisely because it promises to remove work without requiring anyone to change how they think. Point a tool at the tedious task, and the tedium goes away.
It works when the underlying process is sound. When it is not, automation does something worse than failing: it preserves the flaw, accelerates it, and makes it considerably harder to see.
What automation does to a bad process
It removes the humans who were quietly fixing it
Most manual processes work partly because people exercise judgement — noticing the obviously wrong figure, chasing the missing approval. Automate the steps and that correction layer disappears, usually without anyone noticing it was there.
It makes the logic invisible
A manual process can be observed. An automated one becomes a system that people stop interrogating, so its assumptions quietly become facts.
It raises the cost of change
Once a flawed process is encoded, changing it means a change request rather than a conversation.
The boring part, briefly
- Write down what actually happens — not what the documentation says.
- Ask why each step exists. Expect some answers to be “I don't know” or “a regulation that changed years ago”.
- Remove the steps that no longer have a reason.
- Settle disagreements about how it should work, before a tool settles them for you.
- Then automate what remains.
Step two is where the value is. In our experience it is common to find steps that exist because of a system that was replaced, a person who left, or a one-off incident nobody wants to revisit.
The honest version of the promise
Automation does not save time by itself. Removing unnecessary work saves time; automation then makes the remaining necessary work cheaper. Skipping the first part and expecting the second to compensate is the most common way automation projects disappoint.