What are the four questions?

For every assessment answer that places an area at level +2 or above on Di Santo's AIMF scale, the first AI system running inside a business process in production: which system is it, named as you would to a colleague; which business process does it run in, on real inputs with real consequences; how long has it been running, with three months as his reference and no standard behind it; and who would notice within a day if it stopped, with the owner's name.

Why is time in production the misleading question?

Because a system can be paid for and counted for a long time without anyone depending on it. My 12 accounts were on plan for eighteen months by the tracker and would have counted on the long-running side of any survey. Di Santo flags the same weakness in the Gartner figure he reports, 45% of high-maturity organizations keeping projects running three years or more, because both halves of that comparison come from the same self-ratings.

What are the two numbers that verify the fourth question?

Time to landing, the days from signature to the first change the customer makes without you, and reversal rate, the share of go-live workflows the customer switched off, overrode, or routed around by day 60. Both are on the FDE scorecard already. A system someone would miss within a day has a recent change made by someone outside the team that built it and has not been routed around.

What should the forward deployed engineer do in week one?

Run the fourth question on every system the assessment counted, asked of whoever runs each process, because the roadmap and the handshake were both sorted at the declared level. Then take Di Santo's phase 1 gate, the process owner would object if you switched it off, and run it as an action: pick a date and switch one system off for a week. The objection, or the silence, is the level.

How to Check an AI Maturity Self-Assessment

How do you check an AI maturity self-assessment before writing the roadmap?

THE SHORT ANSWER

Ask who would notice within a day if the system stopped, and ask the person who runs the process rather than the leader who filled in the form. Marco Di Santo's October 27, 2026 check puts four questions to every answer claiming a system in production (which system, which process, how long, who would notice in a day), with three months as his reference for how long. The time question is the one that misleads: in my per-seat sunset, 12 accounts sat as on plan for eighteen months, paying committed minimums and using less than the minimum, and nobody would have noticed if it stopped until we lost two of them. An expense agent I put in an approver's seat ran two weeks, approvals rose 40%, about $2,400 was wrongly approved, and finance found out by switching it off. Verify the fourth question with two numbers from the FDE scorecard, the date of the last change the customer made without you and whether the workflow was switched off or routed around by day 60, and then run the gate as an action: pick a date and switch it off for a week.

Ask who would notice within a day if it stopped, and ask the operator, not the leader. The other three of Marco Di Santo's four questions are the address of that one, and the time question is the one that fools a careful person.

The two cases, the two scorecard numbers, and the switch-off week are in Who Would Notice in a Day? The Maturity Level Nobody Checked Has a Test. The numbers come from The FDE Scorecard: Count What Stayed, Not Who Was Sent, and the 12 accounts are in Drawdown Is Not Adoption.

Last reviewed 2026-10-27 · 1 min read

How do you check an AI maturity self-assessment before writing the roadmap?

SOURCES

THE LONG VERSION

RELATED ANSWERS