Multiple teams reported AI-powered advances that directly expand therapeutic design toolkits. Researchers described synthetic, RNA-guided CRISPR-like nucleases designed using AI that rival or outperform natural enzymes, extending the genome-editing toolbox beyond existing sequence families. Separately, Jennifer Doudna’s group reported an AI-guided inverse protein design strategy that created synthetic nucleases with improved activity and reduced off-target activity, using inverse folding models plus evolutionary residue constraints to preserve RNA and DNA compatibility. Together, the papers show how structure-aware models can expand the protein search space while maintaining functional constraints. The same AI expansion theme continues in drug discovery: a Chai-3 release and associated investments emphasize that AI-generated binding improvements in antibody design are now being treated as measurable, experimentally validated outcomes—not just platform promises.