What it was for
Estimation existed to apportion the scarcest resource in the building: engineering hours. Points let teams forecast capacity, negotiate scope, and protect themselves from commitments they could not meet. The ritual's elaborate machinery — poker, velocity, burndown — was a measurement system for a constraint everyone could feel.
The verdict
DISSOLVES. The practice loses its referent. When a working implementation costs hours instead of weeks, the build is cheaper than the forecast of the build — and infinitely more accurate. There is nothing left to estimate that a running system does not answer faster. Velocity charts, the practice's accounting layer, dissolve with it.
What changes
The question "how big is this?" gives way to "is this worth verifying?" — because verification, not construction, is what now consumes the team's scarce attention. Forecasting conversations move up a level, from task-sizing to bet-sizing: which intents deserve a revolution of the engine at all. That is a judgment question, and no deck of cards answers it.
The strongest objection
Enterprises forecast budgets and dependencies on estimates; you cannot run a portfolio on vibes. Conceded — but the honest unit of portfolio forecasting was never the story point. It was the bet, and it still is. Estimation theatre gave that bet false precision; the engine removes the theatre, not the need for judgment.
Sharpened: estimating implementation effort is what dissolves; estimating total change risk — verification load, migration and rollout risk, dependency and portfolio cost — does not. The replacement is bet-sizing against consequence, not the abolition of forecasting.
Falsification: this is wrong wherever sequencing, not construction, is the binding constraint.