The project Traident (Transcriptome-guided AI deconvolution of taxonomy) will develop a novel computational framework for analyzing environmental RNA (eRNA) sequencing data. Unlike DNA, which primarily indicates species presence, RNA provides dynamic information about organisms: their developmental stage, health, stress conditions, and the presence of RNA viruses. Recent discoveries demonstrate that RNA fragments can persist in the environment much longer than expected, making eRNA a powerful yet underutilized resource for biodiversity studies, food safety, museomics, and even clinical research. Existing tools, such as Kraken2, are optimized for DNA and do not account for the specific features of RNA, where highly abundant transcripts like microRNAs and ribosomal RNAs carry rich taxonomic and biological information. Traident will address this gap by integrating genome-wide, microRNA, and rRNA analyses, and applying a deep learning classifier to disentangle complex mixtures of species and estimate their relative contributions. The model will be trained using in silico RNA-seq mixtures from diverse organisms, ensuring robustness and accuracy. Applications of Traident include reconstructing past ecosystems from ancient specimens, monitoring airborne and aquatic biodiversity, detecting food fraud and pathogens in processed products, and uncovering links between disease and microbial presence in clinical cohorts. By providing the first dedicated framework for eRNA deconvolution, this project will open new frontiers in ecology, evolutionary biology, and biomedical research.