Stellar kinematics provides a window into the gravitational field, and therefore into the distribution of all mass, including dark matter. Deep Potential is a non-parametric method for determining both the gravitational potential and the best-fit rotation speed of a system from a snapshot of stellar positions in phase space using techniques borrowed from deep learning. The method has been verified on an N-body simulation and is currently being applied in the Solar vicinity (400pc) using Gaia DR3.