Brain tumor (Brat) is a Drosophila TRIM-NHL protein required for embryogenesis and neural stem cell differentiation. Although structural and biochemical studies established that the Brat NHL domain specifically binds RNA, the in vivo requirement for this activity has not been directly tested. Here, we used structure-guided mutagenesis and genome engineering to determine whether RNA recognition is essential for Brat function during development. The direct interaction between Brats NHL domain and RNA containing Brat Binding Sites (BBS) can be abolished by alanine substitution of three separate residues on the NHL surface. We introduced these point mutations into the endogenous brat locus by CRISPR-mediated Scarless Gene Editing to generate three independent RNA-binding defective mutant (RBDmt) alleles. Complementation tests demonstrated that each allele behaves as a strong loss-of-function mutation: homozygotes and hemizygotes are inviable, and RBDmt alleles fail to complement classical brat null and hypomorphic alleles. Lethal phase analysis revealed death predominantly during late larval and pupal stages, consistent with known brat alleles. Consistent with the namesake brat phenotype, RBDmt larval brains exhibited widespread expression of neuroblast markers and a marked reduction of neuronal differentiation. In embryos, these alleles failed to complement female sterile brat alleles and recapitulated characteristic abdominal segmentation defects. Finally, RT-qPCR showed increased expression of endogenous Brat target mRNAs in mutant larvae, consistent with loss of Brat-mediated repression. Together, these results demonstrate that direct RNA binding is the essential molecular activity of Brat and that post-transcriptional regulation of Brat target mRNAs underlies its criti