The numbers were right. The timing wasn't.
The cohort data is real: agent customers expand, seat customers leave. The board reads it and agrees. The activist reads it and says the company that should own this outcome is one with a longer runway. You're replaced in March. The new CEO sells to Microsoft in September, for less than the 2027 offer would have been.
Box Agent lives on inside Copilot as a content skill. Your name is on the patent.
You saw it. You moved a year late, after the add-on had spent the board's patience. In this war, the cost of the right answer is the year you spent on the comfortable one.
One path that gets here
A · Ship the agent as a per-seat add-onC · Pivot hard: build the omni agent. Kill per-seat for new logosA · Make the case. Show the cohort data: agent customers expand 3x, seat customers churn