Gen AI tools offer unprecedented opportunities, but organizations adopting them often experience an initial dip in productivity before seeing sustained gains. This "productivity J-curve" reflects the growing pains of integrating new systems, reorganizing workflows, and investing in complementary capabilities. To bridge the gap between adoption and measurable impact, some smart companies are taking a disciplined approach: organizational experimentation. By designing targeted experiments and using scientific methods to test, refine, and scale promising solutions, firms such as Siemens, Procter & Gamble, and Google are reducing risk while accelerating learning. The path to gen-AI-driven value is neither quick nor linear, but organizations that invest in experimentation will ultimately navigate uncertainty more successfully and turn potential into real performance gains.