Live call
Is AI-designed protein binders as a service a viable business now?
Blocked on a check since 2026-10-25 · ledger #28
A peer-reviewed crossing exists; no named, funded graveyard yet.
Short answer
Computational protein design used to need screening labs because only a small share of designed binders worked. BindCraft, published in Nature in August 2025, reported one-shot designs with high experimental success rates against many targets, without high-throughput screening. The measured crossing is real; the open check is a named, funded graveyard.
- Killing constraint
- Experimental hit rate of designed binders
E7 Algorithmic step - Crossing
- 2025-08-27
BindCraft published in Nature: one-shot binder design with high experimental success.
When most designs fail, every project needs a screening facility; when most work, design becomes a service.
Graveyard
Asserted, not enumerated: 'the 2000s-2010s computational drug-design cohort'. Named, funded companies are needed before a ripe call.
What changed, dated
- Pacesa et al., 'One-shot design of functional protein binders with BindCraft', Nature. source H