Who killed the AI project?
Eighteen months ago the board approved an AI assistant for the whole organization. There was a kickoff, a license budget, a pilot, a rollout plan. Today four people use it, and three of them work in IT.
The budget is gone, the room has gone quiet, and the board wants a name.
Scroll to the board ↓
0 of 4 suspects filed
The verdict
Nobody in the file did anything unreasonable. Each of the four did the sensible thing from where they sat, which is why the project is dead. The leadership wanted something it could show. The people who would have to use the tool every day were never given a reason to want it. IT and the outside consultant, the two parties in between, did their narrow jobs correctly and went home.
In ten years inside institutions I saw this more than once. A project like this fails because someone introduces a tool and skips the two things that decide whether it lives: whether the people who do the work want it, and whether the organization around them is set up to let them use it.
The tools arrive in weeks. People change how they work over years. If no one is paid to stand in that gap, the pilot becomes a demo, and the demo gets filed under “in consolidation.”
What would have saved it
- One page instead of eighty slides. A decision the board can own, with names in the owner column.
- A pilot chosen with the people who do the work, on their tasks and in their language.
- Training in the week the tool arrives. The harder half of the project, planned and budgeted as such.
- Rules that fit the house instead of a template. Signed before the rollout, not after.
- Somebody who stays until people work differently, and who still answers the phone in month fourteen.
I do the harder half. If this case reads like your building, write to me before month fourteen.
stenreiss@gmail.com