Every year, 10,000 Chilean soap bark trees are harvested to extract QS-21 for vaccines against malaria, shingles, and cancer. Now UC Berkeley's Keasling Lab is using AI to re-engineer plant enzymes that struggle in fermentation tanks.

Scientists at a top-20 pharmaceutical company optimized three late-stage therapeutic peptide programs for a top-20 pharmaceutical partner, each requiring simultaneous optimization of potency, specificity, expression, and thermostability. Cradle generated libraries with a ≥50% success rate in matching all constraints while delivering significant potency improvements, completing in one round what the partner estimated would have required five rounds historically.

Scientists at a top-20 pharmaceutical company optimized a chimeric Staphylococcus aureus vaccine antigen for thermostability in a single round. The best candidate achieved a 2.50 °C increase in melting temperature, completing in one round what the partner estimated would have required seven rounds via historical rational design approaches.

Scientists at a top-50 pharmaceutical partner engineered an IgG for six developability properties (potency, aggregation, nonspecificity, cell binding, immunogenicity, expression) across three rounds. After prior screening and optimization efforts failed to produce candidates meeting both potency and developability criteria, CRADLE-1 delivered 10 successful candidates that advanced to the next pipeline stage.

Scientists at Cradle optimized a haloalkane dehalogenase for thermostability and expression while preserving activity across two rounds. The best candidate achieved a melting temperature increase of +20.0 °C to 65.1 °C, 2.06× improved expression, and 1.29× enhanced activity.

Scientists using Cradle optimized VHHs for polyspecific binding to three distinct snake venom neurotoxins, thermostability, and expression across two rounds. The best candidate achieved binding below 100 pM, below 1 nM, and below 2 nM to the three toxins respectively, with a melting temperature of 76.7 °C and 1.59× improved expression.

Scientists at Cradle optimized a VHH for simultaneous binding to SARS-CoV-2 wild type and Omicron, thermostability, and expression across three rounds. The best candidate achieved a K_D of 186 pM against wild type, 11.4 nM against Omicron, a melting temperature of 70.9 °C, and 1.88× improved expression.

How Cradle’s 12 designs would have ranked top-12 in Adaptyv’s protein design competition 

130 teams competed in Adaptyv Bio’s protein design competition to create novel EGFR-binding proteins. Cradle's platform achieved first place with a binding affinity of 1.21nM, which is 8.2 times stronger than Merck's commercial antibody Cetuximab*.

Cradle just beat or tied first place across all four challenges in a major enzyme engineering benchmark - with zero human intervention required.

After 10 rounds and 1,200 enzyme variants tested, our customer hit diminishing returns—then we nearly tripled their P450 activity in just 3 rounds using their historical data.