Shape Up’s appetite is a value judgment, not an estimate: decide how much team time a problem is worth, then shape scope to fit that creative constraint.
Agents keep that principle, but now you have to budget machine execution and human review separately.
Before the work starts, set:
• Appetite: what the problem is worth in person-weeks
• Machine cap: that appetite in agent-hours, using today’s production factor
• Quality floor: what must be true for the result to count
Example: a 2-person, 6-week appetite = 12 person-weeks = 480 human-equivalent hours.
At a 2× production factor, the machine cap is 240 agent-hours. At a 25% review rate, using the full cap implies about 120 hours of actual human review—not 480 hours of human production.
Then shape scope and execution to fit.
Clear the quality floor with the least human review. Spend more agent time—or add scope—only when the expected gain justifies the extra work and review.
As the machine cap approaches, cut optional work, verify what matters, preserve state, and land.
At the cap, stop. Raising it is a new bet.
Human-equivalent team time is the durable unit. Production factor and review rate are exchange rates for today’s agents. Update them from actual runs.