← Voice corpusEdition 2026-10-10, preview.
XThread

Source item

Open original source ↗
amazing release from the chai team, moving the pharma industry forward from discovery to engineering. 15% binding (!) in the lab from zero-shot AI antibody design model. this two magnitudes better than current industry expectations @joshim5 @jackdent @Mattmcpartlon1 @j_boitreaud 👏 huge kudos to this uniquely talented and tiny team. Some trivia - Josh’s original referral into Embed was from another AI founder, whose recommendation amounted to “I don’t know anything about bio, but that guy is human will embodied” Josh and Jack were CS pset partners at Harvard, ofc Jack and I went on a 3 hour walk the first time we met (talking mostly about database abstractions, codegen) and I hate walks but love smart people Matt is the intersection of A++ researcher and amazing woodworker. Easy pick every time Jacques runs a 2:40 marathon?? Miss having these folks working in the @conviction office @joshim5 @jackdent @Mattmcpartlon1 @j_boitreaud sorry, who builds furniture this beautiful I think the code must be good too https://t.co/LZDohGVAus
3 segments

Segment 1 · characters 0–232 · Original segment ↗

Segment 2 · characters 234–902 · Original segment ↗

Segment 3 · characters 904–1052 · Original segment ↗

Quoted context

These are the quoted author’s words.

We’re excited to introduce Chai-2, a major breakthrough in molecular design. Chai-2 enables zero-shot antibody discovery in a 24-well plate, exceeding previous SOTA by >100x. Thread👇 https://t.co/CnO6vO7VmX

chaidiscovery ↗
Library as of 2026-10-22