Radical Numerics Launches AI Biodefense Model

Radical Numerics, a San Francisco startup founded in 2025 by Stanford researchers, has launched Omnii Defense, an AI model designed to detect both natural and engineered pathogens without an internet connection. Using just a laptop and a USB gene sequencer, the tool assigns each genome a 'threat score' between zero and one, combining pathogenicity likelihood with functional predictions about the sequence.

The founders previously built Evo and Evo-2, open-source DNA foundation models developed during doctoral work at Stanford. Their underlying platform, Omnii, goes beyond DNA to incorporate RNA, proteins, and epigenetic signals. In October, Radical used Omnii to engineer custom bacteriophages targeting a drug-resistant E. coli strain — a treatment administered to a patient under FDA emergency authorization.

That announcement drew scrutiny after Radical failed to disclose a collaboration with AI biotech startup Tabula Bioscience. Harvard geneticist George Church, an adviser to both companies, resigned from his role at Radical following the omission.

Unlike earlier tools such as the federally funded Felix system — which matches sequences against known pathogen libraries — Omnii Defense is built to identify patterns associated with biological function, letting it flag threats even when genetic sequences have been significantly altered. CEO Eric Nguyen frames the approach as a biosecurity arms race analogous to what's already happening in cybersecurity.