JURA builds infrastructure for frontier models to self-improve continuously and reports that its standard training loop captures at least one billion protein-candidate interactions per run.
Variational synthesis uses manufacturing-aware generative models to produce physical DNA samples at scale. Continuous variational synthesis adds pre-training, reward-guided fine-tuning, and adaptation to synthesis hardware.
Sovereign AI models validate lead biologic candidates against priority MDR pathogens; collaboration expands to address additional bacterial threats identified by WRAIR collaborators FOR IMMEDIATE RELEASE BOSTON, MA and HONOLULU, HI – June 1, 2026 – JURA Bio, Inc. ("JURA Bio"
JURA Bio and Koa Bio partnered to discover biologic therapeutics against multidrug-resistant bacterial threats, combining JURA Bio's Sovereign AI platform, threat assessment from Walter Reed Army Institute of Research collaborators, and Koa Bio's candidate-scaling and deployment expertise.
JURA builds foundation models for therapeutic design that drive wet-lab experiments to generate data for model improvement; each design, synthesis, and screening cycle produces proprietary functional signal.
