Title: Steering ultracold reactions through long-lived transient intermediates

Abstract: Controlling the pathways and outcomes of reactions is a broadly pursued goal in chemistry. In gas phase reactions, this is typically achieved by manipulating the properties of the reactants, including their translational energy, orientation, and internal quantum state. In contrast, here we influence the pathway of a reaction via its intermediate complex, which is generally too short-lived to be affected by external processes. In particular, the ultracold preparation of potassium-rubidium (KRb) reactants leads to a long-lived intermediate complex (K$_2$Rb$_2^*$), which allows us to steer the reaction away from its nominal ground-state pathway onto a newly identified excited-state pathway using a laser source at 1064 nm, a wavelength commonly used to confine ultracold molecules. Furthermore, by monitoring the change in the complex population after the sudden removal of the excitation light, we directly measure the lifetime of the complex to be $360 \pm 30$ ns, in agreement with our calculations based on the Rice-Ramsperger-Kassel-Marcus (RRKM) statistical theory. Our results shed light on the origin of the two-body loss widely observed in ultracold molecule experiments. Additionally, the long complex lifetime, coupled with the observed photo-excitation pathway, opens up the possibility to spectroscopically probe the structure of the complex with high resolution, thus elucidating the reaction dynamics.