We show that the shape of the inflationary landscape potential may be constrained by analyzing cosmological data. The quantum fluctuations of fields orthogonal to the inflationary trajectory may have probed the structure of the local landscape potential, inducing non-Gaussianity (NG) in the primordial distribution of the curvature perturbations responsible for the cosmic microwave background (CMB) anisotropies and our Universe's large-scale structure. The resulting type of NG (tomographic NG) is determined by the shape of the landscape potential, and it cannot be fully characterized by 3-or 4-point correlation functions. Here we deduce an expression for the profile of this probability distribution function in terms of the landscape potential, and we show how this can be inverted in order to reconstruct the potential with the help of CMB observations. While current observations do not allow us to infer a significant level of tomographic NG, future surveys may improve the possibility of constraining this class of primordial signatures.Is there any feature about our Universe that would require us to assume primordial non-Gaussian initial conditions? Up until now, cosmic microwave background (CMB) and large-scale structure (LSS) observations are fully consistent with the premise that the primordial curvature perturbations were initially distributed according to a perfectly Gaussian statistics [1,2]. This has favored the simplest models of inflation -single field slowroll inflation-based on the steady evolution of a scalar field driven by a flat potential [3][4][5][6][7]. In these models, the self-interactions of the primordial curvature perturbation lead to tiny non-Gaussianities suppressed by the slow-roll parameters characterizing the evolution of the Hubble expansion rate H, during inflation [8][9][10][11].The confirmation of non-Gaussian initial conditions would help us to decipher certain fundamental aspects about inflation [12][13][14][15]. Indeed, non-Gaussianity (NG) can be generated by nonlinearities affecting the evolution of primordial curvature perturbations (denoted as ζ). These nonlinearities are the result of self-interactions, or interactions with other degrees of freedom, such as isocurvature fields (fields orthogonal to the inflationary trajectory in multifield space). Inevitably, perturbation theory limits the extent to which we can study the emergence of NG, forcing us to focus on the lowest order operators (in terms of field powers) in the ζ Lagrangian. Thus, most of the recent effort devoted to the study of NG has relied on parametrizing it with the bispectrum and trispectrum, the amplitudes of the 3-and 4-point correlation functions of ζ in momentum space. Understanding how different interactions lead to different shapes and runnings of the bispectrum has constituted one of the main programs in the study of inflation [12][13][14][15].It is conceivable, however, that certain classes of interactions may lead to NG deviations that cannot be parametrized just with the bispectrum and/or trispectrum....