Biology has to be able to answer back

Three years ago, as the current wave of AI began moving into biology, we had a clear view of what would be needed next and started building toward it: not only the ability to design better molecules, but the ability to ask causal questions directly in living systems.

We would need to READ those systems, WRITE controlled changes into them and increasingly REASON over what came back, so that in vivo function could become part of the learning loop, not just the final test.

This video shows a milestone in that journey: genetically edited brain immune cells responding side by side to a precise microinjury in the cerebellum of an awake, aged mouse. Their colors distinguish control from edited cells, allowing us to watch both responses live in the same tissue.

Injury is only one query. The same architecture lets us introduce changes through genetics 🧬 , light ⚡ or molecules 💊 delivered directly to the brain, then read function back within minutes or return to the same cells over months.

Aging is a demanding proving ground for a much broader goal: making complex biology and disease easier to test causally. Imagine if we could actually see candidate rejuvenation approaches restore function rather than only change molecular profiles. Or observe the live response to an infection or a novel gene delivery vector. The applications multiply from there.