I present a theory of distributed improvement that extends Agile, Lean, and the Theory of Constraints by demonstrating how to simultaneously improve global and local performance in large systems, especially large organisations. I describe simple, effective rules that allow anyone in an arbitrarily large system to “begin where they are” by first making local improvements without degrading the neighborhood (the next level up in the system). Furthermore, by contributing freed up capacity from local improvements to help improve the neighborhood local improvement efforts align without centralized control. The effects of adopting this approach at all levels are cumulative, and lead to a form of global optimization through distributed collaboration and coordination. My central point is that if we wish to improve the whole system, making a local improvement must not carry an obligation to immediately raise local output. Incentive structures and habits that act otherwise (while widespread) must be changed! From a systemic perspective there are typically wiser ways to exploit the freed up capacity than mindlessly doing more work: principally raising local morale, reserving a portion for future improvement, and assisting at the next level of the system by contributing capacity where it is needed and sharing improvements and innovations. Only in the case where the local improvement has been made at a bottleneck should output be increased. Hence Prager’s law: “After you make an improvement, the last thing you should do is more work.” This approach paves the way to a new generation Agile meta-framework — CSIC: Create Slack, Improve, Contribute — that promises unprecedented productivity gains through compounding improvements, as well as elevated staff morale and insights into how best to deploy coaches and other systems thinkers at all organisational levels. Additional ramifications include changes to executive, manager, and knowledge-worker education, and revision of measurement and incentive design, including staff promotion criteria. Overall, this development is exciting because it points to a balanced way forward that simultaneously improves performance of large systems and improves the quality of life for the people who work within the system.