---
type: "synthesis"
clusters: ["C6", "C2"]
id: "meta-experimentation-operating-mode"
sources: ["cross-day"]
---
Multiple segments prescribe a build-to-learn tempo over grand planning. Spine: [[concept-minimum-viable-ai]], [[concept-build-to-learn]], [[concept-controlled-experimentation-ai]], and half-day prototyping ([[framework-half-day-prototyping]]); experiments are for learning, not validation ([[contrarian-learning-vs-validation]]) ([[cd-experimentation-operating-mode]]). Execution: controlled experiments before headcount decisions ([[action-controlled-experiments]]), Moody's 'sprinting into the fog' ([[quote-sprinting-into-fog]], [[concept-continuous-change-process]]). Commercial: workarounds are customer-funded prototypes ([[contrarian-workarounds-are-prototypes]]); found-time gates exploration ([[framework-curiosity-window-alignment]]). Agentic: test-deploy-learn cycles ([[concept-test-deploy-learn-cycles]]). Futures: stage-gated, zero-based, optionality-preserving bets ([[framework-optimizing-unknown]], [[action-stage-gate-capital]]). The unifying rule: fund reversible learning cheaply, commit heavily only where the destination is clear. Companion: [[meta-how-much-to-bet]], [[meta-anticipatory-layoffs]].