How I Decide What to Automate

Finding the repetitive work worth automating without creating more maintenance.

Automation is worthwhile when it removes frequent, predictable work without creating a more mysterious job in its place. I look for repetition, cost and stability before automating, because a script that needs constant supervision has mostly invented a colleague.

Automation earns its place when it removes repetition without hiding important judgement. I am happy for a script to move files or run checks; I am more cautious when it starts making decisions the team cannot easily inspect.

There is a practical tension underneath this topic: we want enough structure to move confidently, but not so much that the structure becomes the work.

Engineering choices are rarely permanent, but they can make the next choice dramatically easier or harder. The useful question is not whether an approach is perfect. It is whether it leaves the team in a good position when reality changes.

There will always be exceptions. The trick is to make them deliberate exceptions rather than habits nobody remembers choosing.