Beyond 5G communication will require very high data rate and thus will have to deal with GHz RF bandwidth. Conventional heterodyne receivers in the D-band manage multiple non-contiguous bands by using parallelized architecture. This approach is costly in terms of power consumption and silicon area. In this paper, a reconfigurable multiband N-path filter is proposed, allowing Artificial Intelligence (AI) to control the center frequencies, number of bands and bandwidth of each channels thanks to gain coefficients pattern. The AI manages spectrum sensing and shapes the receiver filtering profile by sweeping the filter gain coefficients to the optimize value. This principle is illustrated with the design of a 32-path filter, aggregating three different bands in a 0.1 – 1.6 GHz band, with 21.7mW power consumption.