Finding the repeated work

Start by listing every infrastructure task the team did in the last month, however small. Patterns appear fast: certificate renewals, disk cleanups, environment refreshes, access requests. Each repeated task is either a candidate for automation or for a documented routine.

Turning toil into routines

Not everything deserves automation on day one. A written routine — when it runs, who runs it, what "done" looks like — captures most of the value immediately and makes later automation straightforward, because the steps are already explicit.

What a routine looks like

  • A trigger: a schedule, an alert, or a request.
  • A runbook with exact steps and expected output.
  • A named owner and a fallback.
  • A note field for what surprised you — the seed of automation.

Operational lessons

The win is predictability. Once operations run on routines instead of memory, capacity planning becomes honest, handoffs stop hurting, and automation becomes a series of small, safe steps instead of a big project.