Machine-readable, then machine-run
Compressing the dg2n thesis — the AI operating layer for physical retail — into thirty seconds.
Essays, field notes and occasional experiments. Written slowly.
Compressing the dg2n thesis — the AI operating layer for physical retail — into thirty seconds.
Using diffusion-based surrogates to explore the shell design space — shortening a week-long topology study into minutes.
Monitoring five in-situ shell structures in real time. Temperature, strain, rainfall — and how the data changed the way we pour concrete.
A retrospective on WO2016156915A1: what worked, what didn't, and why the next version is generative.
A study of how LLMs can operate on architectural plans as a structured medium, not just a picture.
After 100,000 immersive experiences on GMetri, a thesis on why the next decade will bring the virtual back to the physical.
Weaving formwork directly at site — what Block Research Group documented in 2015, and what we've learned since.