⚡️ How AI designed a cancer-targeting protein

Hey there,

Most of the AI we talk about is trained on what humanity has already written down. Eli Bixby’s problem is the opposite.

Eli is Head of Machine Learning at Cradle, a protein engineering company whose software is used by pharma, agriculture, and industrial biotech teams. In biology, most of the data needed to design a useful protein hasn’t been collected yet. So Cradle’s models don’t just generate answers. They recommend which experiments a scientist should run next, learn from the results, and repeat. Eli says it usually takes two or three rounds to hit a customer’s targets.

Cradle won a public competition to design a protein that binds tightly to a cancer target, out of 130 teams and more than 1,000 submissions. That result is a good proof point for its technology. What happens after a pharma company buys that technology is the more unusual part.

Watch the full episode on YouTube.

Also available on Spotify & Apple Podcasts.

Ready to move from better prompts to automating real work in Claude?

→ Join our Claude Cohort