As a precursor to biobased chemicals, biomass holds enormous potential for meeting the needs of the circular economy. To get the most out of biomass, a new study proposes borrowing tools from machine learning. During anaerobic fermentation, biomass fuels the growth and proliferation of various microorganisms. These microbes, in turn, form organic molecules that can be processed into specialty chemicals, but the conditions and microbes most conducive to this process aren’t always known. To find out, researchers examined bioreactors designed to form medium-chain carboxylates from xylan and lactate. As expected, reducing the hydraulic retention time, or the average time soluble compounds reside in a bioreactor, boosted the formation of useful medium-chain carboxylates. The key was to identify the organisms responsible for this boost. For that, the team used machine learning models designed to link the change in hydraulic retention time to distinct sequences of microbial DNA. These sequences were found to belong to the genera Olsenella, Lactobacillus, Syntrophococcus and Clostridium IV. The ability to accurately identify bioindicators in this manner holds promise for steering communities of microorganisms to generate desired products, and the generality of this approach means it could be adapted to other systems of microorganisms and the chemical interactions they participate in.