Benchmark releases, methodology essays, and engineering writing — paper-thin, fold-clean. Issue 01 lands with v1.0 of the benchmark.
Three targets, three reasons. The case for a small, defensible benchmark over a sprawling, irreproducible one — and why the field has been measuring the wrong thing for two years.
How we packaged eight protein design tools behind one API — and what we lost, and kept, along the way.
We ran 540 designs and asked which interface metric correlates with downstream wet-lab success. Spoiler: ipSAE wins, by a wider margin than expected.
An honest post-mortem of the misses — what the metrics caught, what they didn't, and the failure modes nobody publishes.
Same address as the waitlist. We email Issue 01 when it ships, then nothing for a month at a time. No tracking pixels, no sequences.