I’ve run two AI readiness sessions across four entities this month.
We expected questions about models, vendors, pilots. That’s not what surfaced.
What surfaced instead: data nobody trusts, processes nobody owns, permissions nobody mapped, workflows that only work because “people know how it’s done.”
AI didn’t create these gaps. It exposed them.
And that distinction matters. Because the gaps were always there – they just didn’t matter as much when humans could fill them silently. When you ask a machine to execute a process, the workarounds don’t transfer. The unspoken rules disappear.
The most confident teams in those sessions were often the least ready.
The ones already getting value weren’t making announcements. They were running disciplined operations long before AI showed up – clean data, clear ownership, executable processes. AI just gave them leverage.
Quietly. Without calling it “AI transformation.”
Before implementing AI, writing SOPs is not enough.
Your operations must be unambiguous, actually followed, and executable without interpretation. If a process still depends on human judgment to fill the gaps, you don’t have a process.
You have tribal knowledge.
AI does not run on tribal knowledge.
So when a board asks: “What’s our AI strategy?” – the real question underneath is: “Are our operations even operable by a machine?”
If not, AI won’t fix them. It will amplify the chaos.
AI readiness isn’t an AI problem. It’s an operations problem.