The best first automation is the one nobody wants to defend in a meeting.
The instinct on day one is to pick something impressive: the workflow with the most steps, the most tools, the biggest number attached to it.
Impressive workflows are also the ones with the most exceptions, the most stakeholders, and the most ways to be subtly wrong. They are a bad first move.
Pick the one nobody defends
The best first automation is dull and frequent: the check that happens every morning, the copy between two tabs, the filing that only matters when it is late. Nobody argues about the rule, because everybody already agrees what it is.
It also produces evidence fast. A week of correct runs on a boring workflow does more for the rollout than a month of debate about an important one.
Then widen
Once one rule is trusted, the second is easier — the allow-list exists, the log habit exists, and the team has seen where the agent stops and asks. The impressive workflow is a much better second or third project than a first.
← All posts