We investigate the low-luminosity supernova SN 2016bkv and its peculiar early-time interaction. For that, we compute radiative transfer models using the CMFGEN code. Because SN 2016bkv shows signs of interaction with material expelled by its progenitor, it offers a great opportunity to constrain the uncertain evolutionary channels leading to low-luminosity supernovae. Our models indicate that the progenitor had a mass-loss rate of (6.0 +- 2.0) x 1e-4 Msun/yr (assuming a velocity of 150 km/s). The surface abundances of the progenitor are consistent with solar contents of He and CNO. If SN 2016bkvs progenitor evolved as a single star, it was an odd red supergiant that did not undergo the expected dredge up for some reason. We propose that the progenitor more likely evolved through binary interaction. One possibility is that the primary star accreted unprocessed material from a companion and avoided further rotational and convective mixing until the SN explosion. Another possibility is a merger with a lower mass star, with the primary remaining with low N abundance until core collapse. Given the available merger models, we can only put a loose constraint on the pre-explosion mass around 10-20 Msun, with lower values being favored based on previous observational constraints from the nebular phase.