Distribution grids require better visibility of emerging household flexibility, yet comprehensive device-level measurements are rarely available. This paper investigates whether a limited set of secure, minute-resolution smart-meter measurements can be transformed into grid-relevant knowledge for a residential low-voltage network. Based on 20 TAF10 measurement streams and a transformer measurement from a live German deployment, the approach first quantifies data availability, latency and electrical plausibility. It then applies an interpretable NNLS/XGBoost-assisted population-reconstruction method to estimate household and controllable-load composition from P-Q features. The reconstructed population is embedded in a minute-resolution network model comprising photovoltaic systems, electric vehicles, heat pumps, batteries and decentralised model-predictive home energy management. The study evaluates whether individually optimised flexibility reduces network stress or creates new coincidence effects. The same field-calibrated flexibility representation is subsequently used as the information layer for an Energy Sharing Community, in which alternative 15-minute allocation rules can be assessed before productive community operation. By connecting measurement validation, population reconstruction, grid-impact assessment and community coordination, the proposed workflow demonstrates how secure smart-meter infrastructure can support distribution-grid planning and local renewable-energy sharing without requiring full appliance-level observability.