In this webinar we describe the development of a highly accurate machine-learned potential (MLP) for Co, enabling simulations of large models of bulk material, surfaces, and nanoclusters over extended time scales across a wide range of temperatures and pressures. While non-magnetic itself, the MLP is trained on several thousand spin-polarized ab initio computations performed using MedeA VASP. The resulting MLP closely reproduces the phonon dispersions of hexagonal close-packed (hcp) and face-centered cubic (fcc) Co, Co surface energies, and the relative stabilities of Co nanoparticles of various shapes. The thermal expansion coefficient and the melting temperature of Co computed with this MLP are close to experimental values. Furthermore, this MLP captures nuanced material properties such as vacancy formation energies on nanoparticle vertices. This accuracy and versatility make the potential suitable for a wide array of applications, including modeling the geometry of Co catalytic surfaces.