Today you can use constraint weights to influence the importance of each constraint on a computed plan. But, since multi-objective optimization is a complex interplay of different trade-offs, it's hard to predict the outcome of such an action/change.

This is a request to make it possible to instead start from defining ideal values for the output metrics of a plan, and then let Timefold figure out what the idea combination of constraint weights is - via hyper-tuning - for that desired outcome.