Our updates
Notes on AI × Science, scientific reasoning models, and what we are building.
-
Building molecular machines to make human biology controllable
Falling DNA synthesis cost and growing model capabilities have changed the economics of lab-in-the-loop protein design; evolution supplies the priors that collapse the search space and the modular functions worth borrowing. What GPT‑4b micro already proved, and what we build on top of it.
-
Building productive scientific intelligence
Why AI × Science is the steepest exponential, why science is structurally ripe for disruption, and how a scientific reasoning layer over frontier models — with a lab in the loop to run the experiments it proposes — turns distributed knowledge into compact world models and monetizable products.