This talk will focus on substrate-mediated nucleation and crystal growth of organic charge-transfer salt crystals and its application towards scalable manufacturing of nanowire sensors. Nanowires are widely recognized as the next generation building block for ultrasensitive and ultrafast chemical detection. Despite the research progress very few nanowire sensors have reached the market due to their manufacturing complexity. We are exploring a simple, low-cost electrocrystallization method to deposit nanowires from a solution droplet at room temperature directly on electronic substrates. We observed a phenomenon of 1-D crystal growth on nucleation seeds of small sizes. While the 1-D nanostructures resemble those produced by vapor-liquid-solid (VLS) or solution-liquid-solid (SLS) methods, their growth mechanisms are completely different. In our case, the small nanoparticle seed size (or radius of curvature) is responsible for the small size and 1-D growth of the organic crystals. We now apply the seed-mediated mechanism for the electrocrystallization of charge-transfer salt nanowires as small as 7 nm in diameter. The organic-based charge-transfer salt crystals are an underutilized class of sensor materials compared to carbon nanotubes (CNTs), conducting polymers, and metal oxides. We are also prototyping a gas sensor for ammonia detection by nanowire electrocrystallization on electrode patterns. Our approach significantly deviates from the current competing technologies, which involve complex, multistep fabrication and surface modification procedures. We have made nanowire sensors using different charge-transfer materials including tetrathiafulvalene charge-transfer salts and tetracyanoplatinate salts. We improve control of the nanowire deposition by applied potential, reactant concentration, and surface pattern. These nanowire materials show partial selectivity towards ammonia, ethanol, and other volatile organic compounds. We show further improvement in selectivity by the construction of a sensor array and machine learning to extra sensor data from the impedance spectra of more than one type of sensors.