Not every task is worth automating
Automation has a cost: it has to be built, and then it has to be maintained. So the question is never "can we automate this?" — almost anything can be. The question is which tasks pay that cost back quickly, and which ones are better left to a person. Getting that order right is most of the value.
The traits of a good first candidate
- Frequent: it happens daily or many times a day, so small savings compound.
- Rule-based: the steps are the same each time and don't need judgment.
- Well-defined inputs: the data comes in a predictable shape, not as a free-text email.
- Low-stakes to check: if it goes wrong, it's easy to catch and correct.
What to leave alone for now
Anything rare, anything that changes every time, and anything where a wrong answer is expensive and hard to notice. Automating a messy process just makes the mess faster. If a task needs cleaning up before a human can do it reliably, it needs cleaning up before a machine can either.
Start small, then follow the pain
The best first automation is one nobody will miss doing — a dull, daily task no one enjoys. Ship it, let people feel the time come back, then follow the next complaint. Automation done this way builds trust instead of anxiety, and the roadmap writes itself from what's actually slowing people down.