There is a moment in almost every AI program where the dashboards look great and the results don’t move. Usage is up. People say the tools are helpful. And the business performs exactly as it did a year ago. That moment has a name: the Stage 2 Ceiling.

Stage 2 of the maturity journey is productivity: AI makes individuals faster. The trap is that individual speed is not organizational advantage. When you give a thousand people a faster way to do the same task, in the same process, reporting to the same structure, you get a thousand slightly-faster people and an unchanged P&L. Faster horses, not a car.

This maps onto what Benedict Evans laid out in his 2026 AI Eats the World analysis: enormous trial, minimal retention. Everyone tries it; the value doesn’t stick. The parallel he draws, and it’s the right one, is electricity. Factories that simply bolted electric motors onto their old steam-era layouts got a modest bump. The productivity revolution came decades later, when they redesigned the factory around what electricity made possible.

Why organizations get stuck here

  • Speed is legible; redesign is hard. You can measure minutes saved. You can’t easily measure a process that should no longer exist.
  • Stage 2 feels like success. Adoption metrics go up and to the right, so nobody sounds the alarm.
  • The work wasn’t changed. Tasks got faster; the shape of the work, who does what, in what order, with what judgment, stayed frozen.

Breaking the ceiling

You break it by moving to Stage 3, context to intelligence: redesigning the work itself around the machine, feeding AI your own context, and rebuilding processes rather than accelerating them. That is unglamorous, and it is where the value gap between leaders and everyone else opens up.

What to take into the room: stop reporting AI adoption. Start reporting whether any process has been redesigned. If the honest answer is “none,” you are at the ceiling, no matter how green the usage charts are.