← sources

Is the real cost of an automation the maintenance, not the build?

source: reddit (reddit:AI_Agents)published 2026-11-09 18:03ingested 2026-11-10 02:36

retained text

Building an automation feels productive. You see it run, it works, and it’s satisfying. But the cost that comes later tends to be less visible: an API changes, a field gets renamed, a trigger stops firing, and suddenly someone has to babysit the thing.

It seems like the real question isn’t “can this be automated?” but “who fixes it when it breaks, and how often will that happen?” Some tasks are better left manual, even if automating them looks easy on paper.

Where do you draw the line? What’s the automation you’d tell someone *not* to build, and why?

document fetched 2026-11-10 00:18 · canonical https://reddit.com/r/AI_Agents/comments/1xrytta/is_the_real_cost_of_an_automation_the_maintenance