Aviv Regev

Head of Research & Early Development at Genentech. Presented at Session II on the lab-in-the-loop architecture.

Key Arguments

• Data, AI, and experimental biology are each necessary and individually insufficient — the unifying primitive is the loop
• Genentech’s transformative medicines (inavolisib, giredestrant, divarasib, fenebrutinib) each hinged on aligning target, molecule, dose, and patient
• The combinatorics of doing that by intuition are absurd: ~37 trillion human cells, ~10^60 drug-like molecules
• Foundation models are bending toward generalization across screens, diseases, and modalities
• Collaboration with Recursion on cross-screen foundation models

Notable Programs

• Autogene cevumeran — personalized neoantigen vaccine with BioNTech for pancreatic cancer; immune responders stayed relapse-free dramatically longer in phase 1