Genetically encoded voltage indicators (GEVIs) are powerful, non-invasive tools for recording action potentials in excitable cells. However, most animal cell types are non-excitable, and yet variations in the membrane potential are biologically relevant in these cells as well. Resolving such small voltage signals demands GEVIs with exceptionally high sensitivity. In this study, we applied structure-guided engineering to the GEVI ASAP3 to generate rEstus, a sensor with optimized brightness, voltage sensitivity, and voltage range. rEstus is most sensitive in the resting voltage range of non-excitable cells, exhibits a 3.6-fold improvement in fast voltage spike detection, and allows for absolute voltage calibration at the single-cell level. Using rEstus, we resolved endogenous voltage fluctuations in several non-excitable cell types and demonstrate that correlation analysis of these optically recorded fluctuations provides an easy, non-invasive, real-time readout of electrical gap-junction coupling. Our work provides greatly enhanced tools and methods for the non-invasive study of electrical signaling in excitable and non-excitable cells.