Stop writing rules. Start directing agents.
For 20 years, a big part of my job was writing rules nobody read. Data definitions. Access policies. Stewardship guides. Written by hand in Excel or on SharePoint. If I was lucky, they got filed in a catalog. Either way, they were left to gather dust.
That job is disappearing. Good riddance. I won't miss it.
The agent took the typing, not the job
Here's the fear every data steward is thinking about right now. If an agent writes the rules faster than I can, what's left for me? Work that took my team a quarter now takes days, so the instinct is to feel replaced.
That's understandable. But we need to think about it differently. The agent didn't take the job. It took the typing.
The steward who used to hand-write access policies now declares the outcome and supervises the agents that enforce it. You stop writing rules. You start setting intent and proving results. You go from author to orchestrator.
This isn't wishful thinking
Alation shipped it this year: outcome-based governance. You declare what you want in plain language. This domain is AI-ready. This data is compliant. Agents interpret it, enforce it, and prove it. A human stays in the loop for the exceptions.
The rule-writer became the outcome-setter.
And the door is wide open. McKinsey's 2026 read: only 1 in 3 enterprises is governance-ready for autonomous agents. The other two-thirds are running agents on controls built for a world where only humans touched the data.
The steward moves up, not out
So the steward role doesn't shrink. It moves up. From the person who documented the rules to the person who owns the outcome the agents answer to. Same instinct. Bigger job. This is an exciting time to be in data governance. People are finally starting to understand what it's really for.
If tomorrow your job was to declare outcomes instead of writing rules, what's the first outcome you'd put an agent on?
And what would you refuse to hand over? Tell me. I read everything.