This page is password protected. Enter the password to read it.
A lean working document. No company is signed and no speaker is listed, which makes this the one case where the candidates are already known: two of my own client deployments, disclosed as such below. Drafted now: the tension, the candidate comparison, the debate armed with cited evidence, and the interview guides. Each section notes what fills in once a company is real.
When the company signs: the interviews establish the real authority line, the overrides that moved it, and the moment the domain expert first disagreed with the system and was right, or wrong.
Same gates as every case in this course, with one that belongs to me.
Lead first. Both are real deployments with domain experts supervising daily.
The choice may also turn on scheduling and on which client says yes first. The comparison is written so either can carry the case.
Written for the aviation lead; swaps cleanly to the clinic if the choice flips.
For eleven years [NAME] had made the call himself: runway by runway, NOTAM by NOTAM, the whole feasibility picture assembled by hand before anyone quoted the trip. Now the engine did it in minutes, seventeen checks, each one green, amber, or red, and his name was still the one on the release. In his first week of supervising instead of doing, he overrode the system twice. The first time he was right, and the team called it proof the human was still essential. The second time he was wrong, and nobody called it anything at all. The question he brought to [MONTH]'s ops review was not whether the engine worked. It was what, exactly, he was still checking, and how he would know when to stop.
When the company signs: the real overrides, the real dates, the decision point the board or ops review actually faced.
Each pole armed with cited evidence now, and with the company's own once the interviews run.
What each will hold. EX A can be drafted from the systems now; the rest wait on the company.
Five to six questions each, tagged to exhibits. Protagonist interview is joint.
The first thing he stopped checking himself and how that felt (EX C), the override he was right about and the one he was wrong about (EX B), what he looks at every morning versus what he trusts blind (EX A), what evidence would make him widen the agent's authority, and what he is teaching the next dispatcher or clinician now that the doing is gone.
A day in the supervision seat as it actually runs (EX A), the last time the system surprised him, whether confirming ever slides into rubber-stamping, honestly, and what the red flag or amber check means to him versus what the manual says it means.
How the authority line got set and by whom (EX A), what the override log gets used for, if anything (EX B), and how a new hire learns supervision when there is no doing left to apprentice on.
What the system was designed to surface versus what its users actually look at, where the monitoring is real and where it is decoration, written openly as the vendor's view and labeled that way in the case packet.
The supervision half of the monitoring article: the arc from doing to overseeing, anchored by a real domain expert with a real override log. Same transcripts, two artifacts, cleared for both uses at once.