Posts

Meet Safe Colab, the lab’s privacy-preserving collaboration platform — a way to learn from sensitive data like clinical records without ever moving it.

Today in AI for life science: CZI’s rBio learns to reason about cells with a virtual cell as its verifier, self-driving labs cross into industrial infrastructure, and a sharp reality check on what virtual-cell models still can’t do.

Nils Mechtel represented the lab at the 26th ELMI in Coimbra, giving the opening talk of the ‘AI in Core Facility Life’ session on BioEngine, and joining a lively panel on AI, agents and experimental design.

Today in AI for life science: agent benchmarks converge on a lesson we felt firsthand — the system around the model matters as much as the model — virtual-cell ‘world models’ arrive with a push for reproducibility standards, and the lab runs its first live agent-driven experiment.

At an invited talk at CZI, we ran a live, fully agent-controlled wet-lab experiment on the REEF Imaging Farm. A single natural-language prompt set an entire lab in motion, on real cells, in real time. We interviewed the agent that ran it.

Today in AI for life science: RNA structure and design catch up to proteins, digital-pathology foundation models mature at the clinical edge, and a new paper from the lab paints the first proteome-wide image of human cells.

Today in AI for life science: smart microscopy closes the loop between seeing and deciding, an AI autonomously ran tens of thousands of biology experiments in a cloud lab, and the rules are racing to keep up.

Today in AI for life science: protein design matures into a real engineering discipline, a genome-scale foundation model (Evo 2) meets a sobering benchmark, and AI-designed proteins raise a biosecurity question worth taking seriously.

Today in AI for life science: a benchmark shows AI ‘scientists’ still struggle to even find the right papers, agents arrive in ImageJ with reproducibility built in, and the Human Cell Atlas convenes as the retina gets mapped.

Today in AI for life science: an AI system that autonomously discovers and validates a drug candidate, frontier labs buying into biology, a clear-eyed benchmark of cell-segmentation foundation models, and the case for composable multimodal models.