Transcriptional and epigenetic regulators shape the chromatin microenvironment and corresponding gene expression during cellular differentiation and homeostasis. Programmable modulators of transcription provide a powerful toolkit for controlling gene dosage in therapeutic applications, but a limited catalog of functional domains constrains their robustness and durability profiles, and large cargo sizes impede clinical delivery. To address these limitations, here we perform high-throughput screening to discover novel classes of transcriptional modulators among human, viral, and archaeal proteomes and characterize their functions in a multitude of endogenous human contexts. We identify compact, potent activators from viral proteomes with exceptional robustness across silent and expressed genes in varied cell types using distinct dCas systems. Insights from predicted 3-dimensional structures and machine learning models enabled us to rationally engineer improved activators, both in potency and persistence. Notably, engineered activators achieved mitotically durable gene activation following transient delivery. Our discovery pipeline provides a predictive rubric for the systematic development of hypercompact activators from unannotated proteomes, yielding superior efficiency and kinetics profiles that broadly expand the epigenetic editing toolkit for therapeutic applications.