Why 'automate everything' is the wrong goal

There's a predictable arc in teams that get good at automation: the first few projects are carefully chosen for high impact, and once the tooling and skills are in place, the bar for what gets automated quietly drops. Anything that can be automated starts feeling like it should be, regardless of how much it actually mattered in the first place.

The cost of this drift isn't visible in any single automation, it's visible in the aggregate: a growing collection of workflows nobody remembers building, each individually cheap to maintain but collectively consuming real attention, and a team that's spent capacity on the automatable rather than the impactful.

The discipline that prevents this is the same one that should have governed the first automation: before building, ask what this actually saves and who actually cares, not just whether the process is repetitive enough to automate. A repetitive process that happens twice a month and takes ten minutes isn't worth the maintenance tax a workflow requires, however satisfying it is to automate.

"Can this be automated" is almost always yes. "Should this be automated" is a much smaller list, and the teams that keep asking the second question stay focused on the automations that actually move something, instead of accumulating infrastructure for its own sake.