Tools vs Products: Lessons learned from a niche industry
summary
AI folding models like AlphaFold2 solve structure quickly but not design end to end so a usable product ties computation to lab cycles and regulatory constraints, not just models. Levitate Bench started work in 2014 as a spin-off from Cyrus after years of contract research and shows customers lack clusters, avoid cloud for IP, and need UX that prevents misinterpretation. All foundation model training data is public so competitors copy models fast; the product gains value by integrating workflows, experiments, and rapid feature delivery.
AI-written summary · about
related talks
36mAlgorithmic Alchemy: Transforming Drug Discovery with Open AI
27mPlatforms for valuable AI Products- Iteration, iteration, iteration
04mHow Machine Learning tools at Google made AlphaFold possible
42mAI-powered products and product-powered AI
53mHow machine learning is influencing protein engineering
14mAI Is Built to Be Applied
01h 13mDeep Learning in Structural Biology and Protein Design: How, Where, and Why
12mKeynote: What's New in AI