Classifying organisms is easier, cheaper, and more accurate than ever thanks to state-of-the-art gene sequencing techniques and powerful machine learning algorithms. Now, researchers report a method that boosts classification resolution beyond the genus level, a well-known barrier faced by 16S rRNA gene sequencing. The principles behind the method are simple: make classification algorithms smarter by training them on richer genetic reference databases. That starts with collecting high-quality genetic datasets for a given habitat, and then identifying reference sequences within these datasets. These sequences help capture the natural genetic variability within each group of organisms in the selected habitat. The team tested their method by creating a training set for a familiar microbial habitat: the aerodigestive tract. Feeding that data into the naïve Bayesian RDP Classifier yielded species-level classification, breaking the genus-level barrier that has traditionally limited 16S rRNA gene microbiome studies. With further refinement, this new approach could help classify and assign sequences to known taxonomical structures.