Newsletter coverage confirms Anthropic published a formal report on Claude's drug discovery applications
Concluded·following since 18 Aug 2026·Day 3·3 sources·updated 19 Aug 2026
Claude autonomous protein binder experiment
The gist
Claude designed validated protein binders for 14 of 15 drug targets at 26.8% hit rate
Designing molecules that bind tightly to drug targets traditionally requires weeks or months of expert work per target. These results show an AI system autonomously completing that step with independently validated wet-lab results competitive with human-led efforts.
The full picture
Anthropic tested whether Claude could autonomously design protein binders from scratch. Given a protocol written by a human expert, Claude handled the full computational pipeline, covering target research, tool installation, candidate generation, filtering, and final selection, without human intervention on individual design decisions. Adaptyv Bio and Twist Bioscience independently synthesized and tested the resulting proteins, finding 354 of 1,320 measured designs bound their intended targets, a 26.8% hit rate, with Claude producing binders for 14 of 15 targets. On 4 of 6 targets comparable to human-led open design competitions, Claude achieved higher hit rates. A Mythos Preview model, given a dedicated 24-hour campaign per target rather than a shared 48-hour campaign across targets, reached a 35.1% hit rate. One identified limitation is that Claude cannot reliably detect when an entire campaign has failed; unsuccessful targets sometimes received computational scores similar to successful ones.
How it developed
Anthropic published results on August 18 showing that Claude, following a protocol written by a human expert, autonomously ran the full computational steps of a protein-design campaign, including target research, tool installation, candidate generation, and final selection, without human decisions on individual designs.
Adaptyv Bio and Twist Bioscience independently built and tested the resulting proteins, finding 354 of 1,320 designs bound their intended targets, a 26.8% hit rate, with Claude producing at least one successful binder for 14 of 15 targets. On 4 of 6 targets comparable to human-led open design competitions, Claude achieved higher hit rates; a Mythos Preview model given a dedicated 24-hour per-target campaign reached 35.1%. One identified limitation: Claude cannot reliably detect when an entire campaign has failed, as unsuccessful targets sometimes received computational scores similar to successful ones.
Details published: 354 of 1,320 measured designs bound targets; Mythos Preview model reached 35.1% hit rate with dedicated per-target campaigns
Sources
Anthropic Twitter
Rohan Paul Twitter
Zvi's AI Roundups
The daily email
Want this in your inbox?
I send a short email each morning with the stories that moved. If you would rather just read here, that works too.
Subscribe free