Most of the talk about AI in marketing is about content. In marketing operations the bigger win is the admin around the work.
Where it earns its place
Messy inputs into clean records
Same notes, same fields, far less retyping. That's where the two hours went.
First drafts of the dull stuff
Field definitions, process docs, release notes. AI gets you most of the way, so your time goes on the part that needs judgement.
Queries and debugging
Describe a report filter or a query in plain English, then check the logic. It beats a blank page, and it's a big part of why I rolled Claude Code out to my team.
Where it doesn't, yet
Bad data
AI won't fix duplicates or a routing rule nobody owns. It just moves bad data faster.
Decisions with consequences
Lead routing, scoring thresholds, who owns which account. These need someone who understands the trade-offs and can explain them to sales.
Anything you can't check
If you can't verify the output, you can't ship it.
Automate the steps where a mistake is cheap and easy to spot. Keep a person on the steps where a mistake is expensive or quiet.
Start with one painful, repetitive task, measure before and after, and only then scale it.
I'll add more as I try things.
