An AI model, termed as LEA (Latent Environment Allocation) has been developed to evaluate the heterogeneity of microbial communities. This study is based on the idea that the microbial ecosystem reflects environments as continuous mixtures. LEA is a machine learning model that predicts the environmental composition of a new sample by placing it on a global microbiome map and comparing it with thousands of existing samples. The model can be applied for environmental monitoring, health diagnostics, and microbiome research, enabling classification of new samples, detection of contamination or dysbiosis, and semantic searching across large datasets without relying on rigid environmental labels.