Manual identification of macroinvertebrates is a pillar of biodiversity observation in freshwater sciences, but it is time-consuming, costly, and relies on taxonomic expertise. To address these limitations, we are developing iNNvert, an AI-powered tool designed to automate the identification, measurement, and enumeration of macroinvertebrates. Our initial dataset, composed of images of benthic samples collected from 20 streams across North America (n = 17 streams from the National Ecological Observatory Network, 3 other streams), offers a suite of macroinvertebrate samples spanning a national scale, ensuring iNNvert’s adaptability across varied contexts. CVAT (Computer Vision Annotation Tool) serves as a preparatory tool facilitating data collection and pixel-level semantic annotation. Images of these macroinvertebrates, taken at standardized distances and different orientations, were annotated using CVAT to generate a robust training dataset for iNNvert. iNNvert performs the automated segmentation, localization, and classification of macroinvertebrates, streamlining data collection, improving efficiency, and reducing dependency on human expertise. Ultimately, iNNvert makes macroinvertebrate identification more accessible and reliable. Initially, we generated a comprehensive annotated dataset, and we have begun using early images and labels for model training. Next, we will implement and refine an advanced YOLO (You Only Look Once) based neural network, an object detection algorithm that will enhance iNNvert’s functionality. The integration of YOLO will allow iNNvert to address challenges such as scale variation and diverse morphological characteristics, making it a versatile tool for ecological monitoring. Preliminary results highlight iNNvert’s potential to classify, measure, and count macroinvertebrates from varied benthic samples. Future directions include validating iNNvert on NEON stream datasets to assess accuracy and refine the model. Additionally, we will explore using iNNvert to enumerate field-collected images of live samples, enabling a novel, low-impact assessment of biodiversity that circumvents the need to sacrifice macroinvertebrates during sampling. In utilizing AI, iNNvert can transform macroinvertebrate identification and biodiversity assessment, paving the way for more reliable and sustainable biodiversity monitoring.