A longstanding problem in galaxy evolution is the lack of a non-arbitrary galaxy size indicator. When digging into the low surface brightness of galaxy disks we realize that there are sudden drops in the surface brightness profiles, dubbed as galaxy truncations, that represent the limit of the radial location of the gas density threshold for star formation. Therefore, they are true galaxy edges that, following the galaxy two-phase formation scenario, characterize up to which distance the in-situ star formation takes place. We compiled ultradeep HST observations of 1048 massive (M_stellar > 10^10 M_Sun) disks to obtain the size evolution for disk galaxies at 0 < z < 1 (Buitrago et al. 2023), also extending our efforts by utilizing Machine Learning algorithms (Fernández-Iglesias et al. 2023, Vega-Ferrero et al. in prep.). Milky Way analogs increase their sizes by a factor of two since z = 1, while at the same time decreasing their stellar surface mass density at the truncation position by an order of magnitude. The size evolution using these galaxy edges/truncations is more pronounced than in the effective radius case, with a factor of two better scatter in the mass-size relation, following a power law of (1+z)^-1. This fact is more in line with the evolution of the size of disk galaxies based on theoretical expectations, as effective radii are driven by light concentration, and thus promotes the role of our promising size proxy for better trace galaxy mass –both baryonic and dark matter– assembly.